2023-05-16 17:45:51

by Sui Jingfeng

[permalink] [raw]
Subject: [PATCH] drm/drm_vblank.c: avoid unsigned int to signed int cast

From: Sui Jingfeng <[email protected]>

Both mode->crtc_htotal and mode->crtc_vtotal are u16 type,
mode->crtc_htotal * mode->crtc_vtotal will results a unsigned type.
Using a u32 is enough to store the result, but considering that the
result will be casted to u64 soon after. We use a u64 type directly.
So there no need to cast it to signed type and cast back then.

Signed-off-by: Sui Jingfeng <[email protected]>
Reviewed-by: Thomas Zimmermann <[email protected]>
Cc: Maarten Lankhorst <[email protected]>
Cc: Maxime Ripard <[email protected]>
Cc: Thomas Zimmermann <[email protected]>
Cc: David Airlie <[email protected]>
Cc: Daniel Vetter <[email protected]>
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
---
drivers/gpu/drm/drm_vblank.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/drm_vblank.c b/drivers/gpu/drm/drm_vblank.c
index 877e2067534f..d99c404b181b 100644
--- a/drivers/gpu/drm/drm_vblank.c
+++ b/drivers/gpu/drm/drm_vblank.c
@@ -622,7 +622,7 @@ void drm_calc_timestamping_constants(struct drm_crtc *crtc,

/* Valid dotclock? */
if (dotclock > 0) {
- int frame_size = mode->crtc_htotal * mode->crtc_vtotal;
+ u64 frame_size = mode->crtc_htotal * mode->crtc_vtotal;

/*
* Convert scanline length in pixels and video
@@ -630,7 +630,7 @@ void drm_calc_timestamping_constants(struct drm_crtc *crtc,
* in nanoseconds:
*/
linedur_ns = div_u64((u64) mode->crtc_htotal * 1000000, dotclock);
- framedur_ns = div_u64((u64) frame_size * 1000000, dotclock);
+ framedur_ns = div_u64(frame_size * 1000000, dotclock);

/*
* Fields of interlaced scanout modes are only half a frame duration.
--
2.25.1



2023-05-17 11:09:14

by David Laight

[permalink] [raw]
Subject: RE: [PATCH] drm/drm_vblank.c: avoid unsigned int to signed int cast

From: [email protected]
> Sent: 16 May 2023 18:30
>
> From: Sui Jingfeng <[email protected]>
>
> Both mode->crtc_htotal and mode->crtc_vtotal are u16 type,
> mode->crtc_htotal * mode->crtc_vtotal will results a unsigned type.

Nope, u16 gets promoted to 'signed int' and the result of the
multiply is also signed.

> Using a u32 is enough to store the result, but considering that the
> result will be casted to u64 soon after. We use a u64 type directly.
> So there no need to cast it to signed type and cast back then.
....
> - int frame_size = mode->crtc_htotal * mode->crtc_vtotal;
> + u64 frame_size = mode->crtc_htotal * mode->crtc_vtotal;
...
> - framedur_ns = div_u64((u64) frame_size * 1000000, dotclock);
> + framedur_ns = div_u64(frame_size * 1000000, dotclock);

The (u64) cast is there to extend the value to 64bits, not
because the original type is signed.
The compiler will detect that the old code is a 32x32 multiply
where a 64bit result is needed, that may not be true for the
changed code (it would need to track back as far as the u16s).

It is not uncommon to force a 64bit result from a multiply
by making the constant 64bit. As in:
div_u64(frame_size * 1000000ULL, dotclock);

David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)


2023-05-17 18:10:50

by Sui Jingfeng

[permalink] [raw]
Subject: Re: [PATCH] drm/drm_vblank.c: avoid unsigned int to signed int cast


On 2023/5/17 18:59, David Laight wrote:
> From: [email protected]
>> Sent: 16 May 2023 18:30
>>
>> From: Sui Jingfeng <[email protected]>
>>
>> Both mode->crtc_htotal and mode->crtc_vtotal are u16 type,
>> mode->crtc_htotal * mode->crtc_vtotal will results a unsigned type.
> Nope, u16 gets promoted to 'signed int' and the result of the
> multiply is also signed.

I believe that signed or unsigned is dependent on the declaration.

I am talk about the math, while you are talking about compiler.

I admit that u16 gets promoted to 'signed int' is true, but this is
irrelevant,

the point is how to understand the returned value.


How does the compiler generate the code is one thing, how do we
interpret the result is another

How does the compiler generate the code is NOT determined by us, while
how do we interpret the result is determined by us.


I believe that using a u32 type to interpret the result(u16 * u16) is
always true, it is true in the perspective of *math*.

Integer promotions is the details of C program language. If the result
of the multiply is signed, then there are risks that

the result is negative, what's the benefit to present this risk to the
programmer?

What's the benefit to tell me(and others) that u16 * u16 yield a signed
value? and can be negative?

Using int type as the return type bring concerns to the programmer and
the user of the function,

even though this is not impossible in practice.

>> Using a u32 is enough to store the result, but considering that the
>> result will be casted to u64 soon after. We use a u64 type directly.
>> So there no need to cast it to signed type and cast back then.
> ....
>> - int frame_size = mode->crtc_htotal * mode->crtc_vtotal;
>> + u64 frame_size = mode->crtc_htotal * mode->crtc_vtotal;
> ...
>> - framedur_ns = div_u64((u64) frame_size * 1000000, dotclock);
>> + framedur_ns = div_u64(frame_size * 1000000, dotclock);
> The (u64) cast is there to extend the value to 64bits, not
> because the original type is signed.

Sorry about my expression, I think my sentence did not mention anything
about 'because the original type is signed'.

In the contrary, my patch eliminated the concerns to the reviewer. It
say that the results of the multiply can't be negative.

My intent is to tell the compiler we want a unsigned return type, but
GCC emit 'imul' instruction for the multiply......

I'm using u64 as the return type, because div_u64() function accept a
u64 type value as its first argument.

> The compiler will detect that the old code is a 32x32 multiply
> where a 64bit result is needed, that may not be true for the
> changed code (it would need to track back as far as the u16s).

I don't believe my code could be wrong.

when you use the word 'may', you are saying that it could be wrong after
apply my patch.

Then you have to find at least one test example to prove you point, in
which case my codes generate wrong results.

Again I don't believe you could find one.

> It is not uncommon to force a 64bit result from a multiply
> by making the constant 64bit. As in:
> div_u64(frame_size * 1000000ULL, dotclock);

In fact, After apply this patch, the ASM code generated is same with before.

This may because the GCC is smart enough to generate optimized code in
either case,

I think It could be different with a different optimization-level.

I have tested this patch on three different architecture,  I can not
find error still.

Below is the assembly extract on x86-64: because GCC generate the same
code in either case,

so I pasted only one copy here.


0000000000000530 <drm_calc_timestamping_constants>:
     530:    f3 0f 1e fa              endbr64
     534:    e8 00 00 00 00           callq  539
<drm_calc_timestamping_constants+0x9>
     539:    55                       push   %rbp
     53a:    48 89 e5                 mov    %rsp,%rbp
     53d:    41 57                    push   %r15
     53f:    41 56                    push   %r14
     541:    41 55                    push   %r13
     543:    41 54                    push   %r12
     545:    53                       push   %rbx
     546:    48 83 ec 18              sub    $0x18,%rsp
     54a:    4c 8b 3f                 mov    (%rdi),%r15
     54d:    41 8b 87 6c 01 00 00     mov    0x16c(%r15),%eax
     554:    85 c0                    test   %eax,%eax
     556:    0f 84 ec 00 00 00        je     648
<drm_calc_timestamping_constants+0x118>
     55c:    44 8b 87 90 00 00 00     mov    0x90(%rdi),%r8d
     563:    49 89 fc                 mov    %rdi,%r12
     566:    44 39 c0                 cmp    %r8d,%eax
     569:    0f 86 40 01 00 00        jbe    6af
<drm_calc_timestamping_constants+0x17f>
     56f:    44 8b 76 1c              mov    0x1c(%rsi),%r14d
     573:    49 8b 8f 40 01 00 00     mov    0x140(%r15),%rcx
     57a:    48 89 f3                 mov    %rsi,%rbx
     57d:    45 85 f6                 test   %r14d,%r14d
     580:    0f 8e d5 00 00 00        jle    65b
<drm_calc_timestamping_constants+0x12b>
     586:    0f b7 43 2a              movzwl 0x2a(%rbx),%eax
     58a:    49 63 f6                 movslq %r14d,%rsi
     58d:    31 d2                    xor    %edx,%edx
     58f:    48 89 c7                 mov    %rax,%rdi
     592:    48 69 c0 40 42 0f 00     imul   $0xf4240,%rax,%rax
     599:    48 f7 f6                 div    %rsi
     59c:    31 d2                    xor    %edx,%edx
     59e:    48 89 45 d0              mov    %rax,-0x30(%rbp)
     5a2:    0f b7 43 38              movzwl 0x38(%rbx),%eax
     5a6:    0f af c7                 imul   %edi,%eax
     5a9:    48 98                    cltq
     5ab:    48 69 c0 40 42 0f 00     imul   $0xf4240,%rax,%rax
     5b2:    48 f7 f6                 div    %rsi
     5b5:    41 89 c5                 mov    %eax,%r13d
     5b8:    f6 43 18 10              testb  $0x10,0x18(%rbx)
     5bc:    74 0a                    je     5c8
<drm_calc_timestamping_constants+0x98>
     5be:    41 c1 ed 1f              shr    $0x1f,%r13d
     5c2:    41 01 c5                 add    %eax,%r13d
     5c5:    41 d1 fd                 sar    %r13d
     5c8:    4b 8d 04 c0              lea    (%r8,%r8,8),%rax
     5cc:    48 89 de                 mov    %rbx,%rsi
     5cf:    49 8d 3c 40              lea    (%r8,%rax,2),%rdi
     5d3:    8b 45 d0                 mov    -0x30(%rbp),%eax
     5d6:    48 c1 e7 04              shl    $0x4,%rdi
     5da:    48 01 cf                 add    %rcx,%rdi
     5dd:    89 47 78                 mov    %eax,0x78(%rdi)
     5e0:    48 83 ef 80              sub $0xffffffffffffff80,%rdi
     5e4:    44 89 6f f4              mov    %r13d,-0xc(%rdi)
     5e8:    e8 00 00 00 00           callq  5ed
<drm_calc_timestamping_constants+0xbd>
     5ed:    0f b7 53 2e              movzwl 0x2e(%rbx),%edx
     5f1:    0f b7 43 38              movzwl 0x38(%rbx),%eax
     5f5:    44 0f b7 4b 2a           movzwl 0x2a(%rbx),%r9d
     5fa:    45 8b 44 24 60           mov    0x60(%r12),%r8d
     5ff:    4d 85 ff                 test   %r15,%r15
     602:    0f 84 87 00 00 00        je     68f
<drm_calc_timestamping_constants+0x15f>
     608:    49 8b 77 08              mov    0x8(%r15),%rsi
     60c:    52                       push   %rdx
     60d:    31 ff                    xor    %edi,%edi
     60f:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
     616:    50                       push   %rax
     617:    31 d2                    xor    %edx,%edx
     619:    e8 00 00 00 00           callq  61e
<drm_calc_timestamping_constants+0xee>
     61e:    45 8b 44 24 60           mov    0x60(%r12),%r8d
     623:    4d 8b 7f 08              mov    0x8(%r15),%r15
     627:    5f                       pop    %rdi
     628:    41 59                    pop    %r9
     62a:    8b 45 d0                 mov    -0x30(%rbp),%eax
     62d:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
     634:    4c 89 fe                 mov    %r15,%rsi
     637:    45 89 f1                 mov    %r14d,%r9d
     63a:    31 d2                    xor    %edx,%edx
     63c:    31 ff                    xor    %edi,%edi
     63e:    50                       push   %rax
     63f:    41 55                    push   %r13
     641:    e8 00 00 00 00           callq  646
<drm_calc_timestamping_constants+0x116>
     646:    59                       pop    %rcx
     647:    5e                       pop    %rsi
     648:    48 8d 65 d8              lea    -0x28(%rbp),%rsp
     64c:    5b                       pop    %rbx
     64d:    41 5c                    pop    %r12
     64f:    41 5d                    pop    %r13
     651:    41 5e                    pop    %r14
     653:    41 5f                    pop    %r15
     655:    5d                       pop    %rbp
     656:    e9 00 00 00 00           jmpq   65b
<drm_calc_timestamping_constants+0x12b>
     65b:    41 8b 54 24 60           mov    0x60(%r12),%edx
     660:    49 8b 7f 08              mov    0x8(%r15),%rdi
     664:    44 89 45 c4              mov    %r8d,-0x3c(%rbp)
     668:    45 31 ed                 xor    %r13d,%r13d
     66b:    48 c7 c6 00 00 00 00     mov    $0x0,%rsi
     672:    48 89 4d c8              mov    %rcx,-0x38(%rbp)
     676:    e8 00 00 00 00           callq  67b
<drm_calc_timestamping_constants+0x14b>
     67b:    c7 45 d0 00 00 00 00     movl   $0x0,-0x30(%rbp)
     682:    44 8b 45 c4              mov    -0x3c(%rbp),%r8d
     686:    48 8b 4d c8              mov    -0x38(%rbp),%rcx
     68a:    e9 39 ff ff ff           jmpq   5c8
<drm_calc_timestamping_constants+0x98>
     68f:    52                       push   %rdx
     690:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
     697:    31 d2                    xor    %edx,%edx
     699:    31 f6                    xor    %esi,%esi
     69b:    50                       push   %rax
     69c:    31 ff                    xor    %edi,%edi
     69e:    e8 00 00 00 00           callq  6a3
<drm_calc_timestamping_constants+0x173>
     6a3:    45 8b 44 24 60           mov    0x60(%r12),%r8d
     6a8:    58                       pop    %rax
     6a9:    5a                       pop    %rdx
     6aa:    e9 7b ff ff ff           jmpq   62a
<drm_calc_timestamping_constants+0xfa>
     6af:    49 8b 7f 08              mov    0x8(%r15),%rdi
     6b3:    4c 8b 67 50              mov    0x50(%rdi),%r12
     6b7:    4d 85 e4                 test   %r12,%r12
     6ba:    74 25                    je     6e1
<drm_calc_timestamping_constants+0x1b1>
     6bc:    e8 00 00 00 00           callq  6c1
<drm_calc_timestamping_constants+0x191>
     6c1:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
     6c8:    4c 89 e2                 mov    %r12,%rdx
     6cb:    48 c7 c7 00 00 00 00     mov    $0x0,%rdi
     6d2:    48 89 c6                 mov    %rax,%rsi
     6d5:    e8 00 00 00 00           callq  6da
<drm_calc_timestamping_constants+0x1aa>
     6da:    0f 0b                    ud2
     6dc:    e9 67 ff ff ff           jmpq   648
<drm_calc_timestamping_constants+0x118>
     6e1:    4c 8b 27                 mov    (%rdi),%r12
     6e4:    eb d6                    jmp    6bc
<drm_calc_timestamping_constants+0x18c>
     6e6:    66 2e 0f 1f 84 00 00     nopw   %cs:0x0(%rax,%rax,1)
     6ed:    00 00 00
     6f0:    90                       nop
     6f1:    90                       nop
     6f2:    90                       nop
     6f3:    90                       nop
     6f4:    90                       nop
     6f5:    90                       nop
     6f6:    90                       nop
     6f7:    90                       nop
     6f8:    90                       nop
     6f9:    90                       nop
     6fa:    90                       nop
     6fb:    90                       nop
     6fc:    90                       nop
     6fd:    90                       nop
     6fe:    90                       nop
     6ff:    90                       nop


> David
>
> -
> Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
> Registration No: 1397386 (Wales)
>

2023-05-22 11:51:21

by Jani Nikula

[permalink] [raw]
Subject: Re: [PATCH] drm/drm_vblank.c: avoid unsigned int to signed int cast

On Thu, 18 May 2023, Sui Jingfeng <[email protected]> wrote:
> On 2023/5/17 18:59, David Laight wrote:
>> From: [email protected]
>>> Sent: 16 May 2023 18:30
>>>
>>> From: Sui Jingfeng <[email protected]>
>>>
>>> Both mode->crtc_htotal and mode->crtc_vtotal are u16 type,
>>> mode->crtc_htotal * mode->crtc_vtotal will results a unsigned type.
>> Nope, u16 gets promoted to 'signed int' and the result of the
>> multiply is also signed.
>
> I believe that signed or unsigned is dependent on the declaration.
>
> I am talk about the math, while you are talking about compiler.
>
> I admit that u16 gets promoted to 'signed int' is true, but this is
> irrelevant,
>
> the point is how to understand the returned value.
>
>
> How does the compiler generate the code is one thing, how do we
> interpret the result is another
>
> How does the compiler generate the code is NOT determined by us, while
> how do we interpret the result is determined by us.
>
>
> I believe that using a u32 type to interpret the result(u16 * u16) is
> always true, it is true in the perspective of *math*.
>
> Integer promotions is the details of C program language. If the result
> of the multiply is signed, then there are risks that
>
> the result is negative, what's the benefit to present this risk to the
> programmer?
>
> What's the benefit to tell me(and others) that u16 * u16 yield a signed
> value? and can be negative?
>
> Using int type as the return type bring concerns to the programmer and
> the user of the function,
>
> even though this is not impossible in practice.

In general, do not use unsigned types in arithmethic to avoid negative
values, because most people will be tripped over by integer promotion
rules, and you'll get negative values anyway.

I'll bet most people will be surprised to see what this prints:

#include <stdio.h>
#include <stdint.h>

int main(void)
{
uint16_t x = 0xffff;
uint16_t y = 0xffff;
uint64_t z = x * y;

printf("0x%016lx\n", z);
printf("%ld\n", z);
printf("%d\n", x * y);
}

And it's not that different from what you have below. Your patch doesn't
change anything, and doesn't make it any less confusing.

BR,
Jani.


>
>>> Using a u32 is enough to store the result, but considering that the
>>> result will be casted to u64 soon after. We use a u64 type directly.
>>> So there no need to cast it to signed type and cast back then.
>> ....
>>> - int frame_size = mode->crtc_htotal * mode->crtc_vtotal;
>>> + u64 frame_size = mode->crtc_htotal * mode->crtc_vtotal;
>> ...
>>> - framedur_ns = div_u64((u64) frame_size * 1000000, dotclock);
>>> + framedur_ns = div_u64(frame_size * 1000000, dotclock);
>> The (u64) cast is there to extend the value to 64bits, not
>> because the original type is signed.
>
> Sorry about my expression, I think my sentence did not mention anything
> about 'because the original type is signed'.
>
> In the contrary, my patch eliminated the concerns to the reviewer. It
> say that the results of the multiply can't be negative.
>
> My intent is to tell the compiler we want a unsigned return type, but
> GCC emit 'imul' instruction for the multiply......
>
> I'm using u64 as the return type, because div_u64() function accept a
> u64 type value as its first argument.
>
>> The compiler will detect that the old code is a 32x32 multiply
>> where a 64bit result is needed, that may not be true for the
>> changed code (it would need to track back as far as the u16s).
>
> I don't believe my code could be wrong.
>
> when you use the word 'may', you are saying that it could be wrong after
> apply my patch.
>
> Then you have to find at least one test example to prove you point, in
> which case my codes generate wrong results.
>
> Again I don't believe you could find one.
>
>> It is not uncommon to force a 64bit result from a multiply
>> by making the constant 64bit. As in:
>> div_u64(frame_size * 1000000ULL, dotclock);
>
> In fact, After apply this patch, the ASM code generated is same with before.
>
> This may because the GCC is smart enough to generate optimized code in
> either case,
>
> I think It could be different with a different optimization-level.
>
> I have tested this patch on three different architecture,  I can not
> find error still.
>
> Below is the assembly extract on x86-64: because GCC generate the same
> code in either case,
>
> so I pasted only one copy here.
>
>
> 0000000000000530 <drm_calc_timestamping_constants>:
>      530:    f3 0f 1e fa              endbr64
>      534:    e8 00 00 00 00           callq  539
> <drm_calc_timestamping_constants+0x9>
>      539:    55                       push   %rbp
>      53a:    48 89 e5                 mov    %rsp,%rbp
>      53d:    41 57                    push   %r15
>      53f:    41 56                    push   %r14
>      541:    41 55                    push   %r13
>      543:    41 54                    push   %r12
>      545:    53                       push   %rbx
>      546:    48 83 ec 18              sub    $0x18,%rsp
>      54a:    4c 8b 3f                 mov    (%rdi),%r15
>      54d:    41 8b 87 6c 01 00 00     mov    0x16c(%r15),%eax
>      554:    85 c0                    test   %eax,%eax
>      556:    0f 84 ec 00 00 00        je     648
> <drm_calc_timestamping_constants+0x118>
>      55c:    44 8b 87 90 00 00 00     mov    0x90(%rdi),%r8d
>      563:    49 89 fc                 mov    %rdi,%r12
>      566:    44 39 c0                 cmp    %r8d,%eax
>      569:    0f 86 40 01 00 00        jbe    6af
> <drm_calc_timestamping_constants+0x17f>
>      56f:    44 8b 76 1c              mov    0x1c(%rsi),%r14d
>      573:    49 8b 8f 40 01 00 00     mov    0x140(%r15),%rcx
>      57a:    48 89 f3                 mov    %rsi,%rbx
>      57d:    45 85 f6                 test   %r14d,%r14d
>      580:    0f 8e d5 00 00 00        jle    65b
> <drm_calc_timestamping_constants+0x12b>
>      586:    0f b7 43 2a              movzwl 0x2a(%rbx),%eax
>      58a:    49 63 f6                 movslq %r14d,%rsi
>      58d:    31 d2                    xor    %edx,%edx
>      58f:    48 89 c7                 mov    %rax,%rdi
>      592:    48 69 c0 40 42 0f 00     imul   $0xf4240,%rax,%rax
>      599:    48 f7 f6                 div    %rsi
>      59c:    31 d2                    xor    %edx,%edx
>      59e:    48 89 45 d0              mov    %rax,-0x30(%rbp)
>      5a2:    0f b7 43 38              movzwl 0x38(%rbx),%eax
>      5a6:    0f af c7                 imul   %edi,%eax
>      5a9:    48 98                    cltq
>      5ab:    48 69 c0 40 42 0f 00     imul   $0xf4240,%rax,%rax
>      5b2:    48 f7 f6                 div    %rsi
>      5b5:    41 89 c5                 mov    %eax,%r13d
>      5b8:    f6 43 18 10              testb  $0x10,0x18(%rbx)
>      5bc:    74 0a                    je     5c8
> <drm_calc_timestamping_constants+0x98>
>      5be:    41 c1 ed 1f              shr    $0x1f,%r13d
>      5c2:    41 01 c5                 add    %eax,%r13d
>      5c5:    41 d1 fd                 sar    %r13d
>      5c8:    4b 8d 04 c0              lea    (%r8,%r8,8),%rax
>      5cc:    48 89 de                 mov    %rbx,%rsi
>      5cf:    49 8d 3c 40              lea    (%r8,%rax,2),%rdi
>      5d3:    8b 45 d0                 mov    -0x30(%rbp),%eax
>      5d6:    48 c1 e7 04              shl    $0x4,%rdi
>      5da:    48 01 cf                 add    %rcx,%rdi
>      5dd:    89 47 78                 mov    %eax,0x78(%rdi)
>      5e0:    48 83 ef 80              sub $0xffffffffffffff80,%rdi
>      5e4:    44 89 6f f4              mov    %r13d,-0xc(%rdi)
>      5e8:    e8 00 00 00 00           callq  5ed
> <drm_calc_timestamping_constants+0xbd>
>      5ed:    0f b7 53 2e              movzwl 0x2e(%rbx),%edx
>      5f1:    0f b7 43 38              movzwl 0x38(%rbx),%eax
>      5f5:    44 0f b7 4b 2a           movzwl 0x2a(%rbx),%r9d
>      5fa:    45 8b 44 24 60           mov    0x60(%r12),%r8d
>      5ff:    4d 85 ff                 test   %r15,%r15
>      602:    0f 84 87 00 00 00        je     68f
> <drm_calc_timestamping_constants+0x15f>
>      608:    49 8b 77 08              mov    0x8(%r15),%rsi
>      60c:    52                       push   %rdx
>      60d:    31 ff                    xor    %edi,%edi
>      60f:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>      616:    50                       push   %rax
>      617:    31 d2                    xor    %edx,%edx
>      619:    e8 00 00 00 00           callq  61e
> <drm_calc_timestamping_constants+0xee>
>      61e:    45 8b 44 24 60           mov    0x60(%r12),%r8d
>      623:    4d 8b 7f 08              mov    0x8(%r15),%r15
>      627:    5f                       pop    %rdi
>      628:    41 59                    pop    %r9
>      62a:    8b 45 d0                 mov    -0x30(%rbp),%eax
>      62d:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>      634:    4c 89 fe                 mov    %r15,%rsi
>      637:    45 89 f1                 mov    %r14d,%r9d
>      63a:    31 d2                    xor    %edx,%edx
>      63c:    31 ff                    xor    %edi,%edi
>      63e:    50                       push   %rax
>      63f:    41 55                    push   %r13
>      641:    e8 00 00 00 00           callq  646
> <drm_calc_timestamping_constants+0x116>
>      646:    59                       pop    %rcx
>      647:    5e                       pop    %rsi
>      648:    48 8d 65 d8              lea    -0x28(%rbp),%rsp
>      64c:    5b                       pop    %rbx
>      64d:    41 5c                    pop    %r12
>      64f:    41 5d                    pop    %r13
>      651:    41 5e                    pop    %r14
>      653:    41 5f                    pop    %r15
>      655:    5d                       pop    %rbp
>      656:    e9 00 00 00 00           jmpq   65b
> <drm_calc_timestamping_constants+0x12b>
>      65b:    41 8b 54 24 60           mov    0x60(%r12),%edx
>      660:    49 8b 7f 08              mov    0x8(%r15),%rdi
>      664:    44 89 45 c4              mov    %r8d,-0x3c(%rbp)
>      668:    45 31 ed                 xor    %r13d,%r13d
>      66b:    48 c7 c6 00 00 00 00     mov    $0x0,%rsi
>      672:    48 89 4d c8              mov    %rcx,-0x38(%rbp)
>      676:    e8 00 00 00 00           callq  67b
> <drm_calc_timestamping_constants+0x14b>
>      67b:    c7 45 d0 00 00 00 00     movl   $0x0,-0x30(%rbp)
>      682:    44 8b 45 c4              mov    -0x3c(%rbp),%r8d
>      686:    48 8b 4d c8              mov    -0x38(%rbp),%rcx
>      68a:    e9 39 ff ff ff           jmpq   5c8
> <drm_calc_timestamping_constants+0x98>
>      68f:    52                       push   %rdx
>      690:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>      697:    31 d2                    xor    %edx,%edx
>      699:    31 f6                    xor    %esi,%esi
>      69b:    50                       push   %rax
>      69c:    31 ff                    xor    %edi,%edi
>      69e:    e8 00 00 00 00           callq  6a3
> <drm_calc_timestamping_constants+0x173>
>      6a3:    45 8b 44 24 60           mov    0x60(%r12),%r8d
>      6a8:    58                       pop    %rax
>      6a9:    5a                       pop    %rdx
>      6aa:    e9 7b ff ff ff           jmpq   62a
> <drm_calc_timestamping_constants+0xfa>
>      6af:    49 8b 7f 08              mov    0x8(%r15),%rdi
>      6b3:    4c 8b 67 50              mov    0x50(%rdi),%r12
>      6b7:    4d 85 e4                 test   %r12,%r12
>      6ba:    74 25                    je     6e1
> <drm_calc_timestamping_constants+0x1b1>
>      6bc:    e8 00 00 00 00           callq  6c1
> <drm_calc_timestamping_constants+0x191>
>      6c1:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>      6c8:    4c 89 e2                 mov    %r12,%rdx
>      6cb:    48 c7 c7 00 00 00 00     mov    $0x0,%rdi
>      6d2:    48 89 c6                 mov    %rax,%rsi
>      6d5:    e8 00 00 00 00           callq  6da
> <drm_calc_timestamping_constants+0x1aa>
>      6da:    0f 0b                    ud2
>      6dc:    e9 67 ff ff ff           jmpq   648
> <drm_calc_timestamping_constants+0x118>
>      6e1:    4c 8b 27                 mov    (%rdi),%r12
>      6e4:    eb d6                    jmp    6bc
> <drm_calc_timestamping_constants+0x18c>
>      6e6:    66 2e 0f 1f 84 00 00     nopw   %cs:0x0(%rax,%rax,1)
>      6ed:    00 00 00
>      6f0:    90                       nop
>      6f1:    90                       nop
>      6f2:    90                       nop
>      6f3:    90                       nop
>      6f4:    90                       nop
>      6f5:    90                       nop
>      6f6:    90                       nop
>      6f7:    90                       nop
>      6f8:    90                       nop
>      6f9:    90                       nop
>      6fa:    90                       nop
>      6fb:    90                       nop
>      6fc:    90                       nop
>      6fd:    90                       nop
>      6fe:    90                       nop
>      6ff:    90                       nop
>
>
>> David
>>
>> -
>> Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
>> Registration No: 1397386 (Wales)
>>

--
Jani Nikula, Intel Open Source Graphics Center

2023-05-22 12:09:24

by Sui Jingfeng

[permalink] [raw]
Subject: Re: [PATCH] drm/drm_vblank.c: avoid unsigned int to signed int cast

Hi,

On 2023/5/22 19:29, Jani Nikula wrote:
> On Thu, 18 May 2023, Sui Jingfeng <[email protected]> wrote:
>> On 2023/5/17 18:59, David Laight wrote:
>>> From: [email protected]
>>>> Sent: 16 May 2023 18:30
>>>>
>>>> From: Sui Jingfeng <[email protected]>
>>>>
>>>> Both mode->crtc_htotal and mode->crtc_vtotal are u16 type,
>>>> mode->crtc_htotal * mode->crtc_vtotal will results a unsigned type.
>>> Nope, u16 gets promoted to 'signed int' and the result of the
>>> multiply is also signed.
>> I believe that signed or unsigned is dependent on the declaration.
>>
>> I am talk about the math, while you are talking about compiler.
>>
>> I admit that u16 gets promoted to 'signed int' is true, but this is
>> irrelevant,
>>
>> the point is how to understand the returned value.
>>
>>
>> How does the compiler generate the code is one thing, how do we
>> interpret the result is another
>>
>> How does the compiler generate the code is NOT determined by us, while
>> how do we interpret the result is determined by us.
>>
>>
>> I believe that using a u32 type to interpret the result(u16 * u16) is
>> always true, it is true in the perspective of *math*.
>>
>> Integer promotions is the details of C program language. If the result
>> of the multiply is signed, then there are risks that
>>
>> the result is negative, what's the benefit to present this risk to the
>> programmer?
>>
>> What's the benefit to tell me(and others) that u16 * u16 yield a signed
>> value? and can be negative?
>>
>> Using int type as the return type bring concerns to the programmer and
>> the user of the function,
>>
>> even though this is not impossible in practice.
> In general, do not use unsigned types in arithmethic to avoid negative
> values, because most people will be tripped over by integer promotion
> rules, and you'll get negative values anyway.
>
> I'll bet most people will be surprised to see what this prints:
>
> #include <stdio.h>
> #include <stdint.h>
>
> int main(void)
> {
> uint16_t x = 0xffff;
> uint16_t y = 0xffff;
> uint64_t z = x * y;
>
> printf("0x%016lx\n", z);
> printf("%ld\n", z);

Here, please replace the "%ld\n" with the "%lu\n", then you will see the
difference.

you are casting the variable 'z' to signed value,  "%d" is for printing
signed value, and "%u" is for printing unsigned value.


Your simple code explained exactly why you are still in confusion,

that is u16 * u16  can yield a negative value if you use the int as the
return type. Because it overflowed.

> printf("%d\n", x * y);
> }
>
> And it's not that different from what you have below. Your patch doesn't
> change anything, and doesn't make it any less confusing.
>
> BR,
> Jani.
>
>
>>>> Using a u32 is enough to store the result, but considering that the
>>>> result will be casted to u64 soon after. We use a u64 type directly.
>>>> So there no need to cast it to signed type and cast back then.
>>> ....
>>>> - int frame_size = mode->crtc_htotal * mode->crtc_vtotal;
>>>> + u64 frame_size = mode->crtc_htotal * mode->crtc_vtotal;
>>> ...
>>>> - framedur_ns = div_u64((u64) frame_size * 1000000, dotclock);
>>>> + framedur_ns = div_u64(frame_size * 1000000, dotclock);
>>> The (u64) cast is there to extend the value to 64bits, not
>>> because the original type is signed.
>> Sorry about my expression, I think my sentence did not mention anything
>> about 'because the original type is signed'.
>>
>> In the contrary, my patch eliminated the concerns to the reviewer. It
>> say that the results of the multiply can't be negative.
>>
>> My intent is to tell the compiler we want a unsigned return type, but
>> GCC emit 'imul' instruction for the multiply......
>>
>> I'm using u64 as the return type, because div_u64() function accept a
>> u64 type value as its first argument.
>>
>>> The compiler will detect that the old code is a 32x32 multiply
>>> where a 64bit result is needed, that may not be true for the
>>> changed code (it would need to track back as far as the u16s).
>> I don't believe my code could be wrong.
>>
>> when you use the word 'may', you are saying that it could be wrong after
>> apply my patch.
>>
>> Then you have to find at least one test example to prove you point, in
>> which case my codes generate wrong results.
>>
>> Again I don't believe you could find one.
>>
>>> It is not uncommon to force a 64bit result from a multiply
>>> by making the constant 64bit. As in:
>>> div_u64(frame_size * 1000000ULL, dotclock);
>> In fact, After apply this patch, the ASM code generated is same with before.
>>
>> This may because the GCC is smart enough to generate optimized code in
>> either case,
>>
>> I think It could be different with a different optimization-level.
>>
>> I have tested this patch on three different architecture,  I can not
>> find error still.
>>
>> Below is the assembly extract on x86-64: because GCC generate the same
>> code in either case,
>>
>> so I pasted only one copy here.
>>
>>
>> 0000000000000530 <drm_calc_timestamping_constants>:
>>      530:    f3 0f 1e fa              endbr64
>>      534:    e8 00 00 00 00           callq  539
>> <drm_calc_timestamping_constants+0x9>
>>      539:    55                       push   %rbp
>>      53a:    48 89 e5                 mov    %rsp,%rbp
>>      53d:    41 57                    push   %r15
>>      53f:    41 56                    push   %r14
>>      541:    41 55                    push   %r13
>>      543:    41 54                    push   %r12
>>      545:    53                       push   %rbx
>>      546:    48 83 ec 18              sub    $0x18,%rsp
>>      54a:    4c 8b 3f                 mov    (%rdi),%r15
>>      54d:    41 8b 87 6c 01 00 00     mov    0x16c(%r15),%eax
>>      554:    85 c0                    test   %eax,%eax
>>      556:    0f 84 ec 00 00 00        je     648
>> <drm_calc_timestamping_constants+0x118>
>>      55c:    44 8b 87 90 00 00 00     mov    0x90(%rdi),%r8d
>>      563:    49 89 fc                 mov    %rdi,%r12
>>      566:    44 39 c0                 cmp    %r8d,%eax
>>      569:    0f 86 40 01 00 00        jbe    6af
>> <drm_calc_timestamping_constants+0x17f>
>>      56f:    44 8b 76 1c              mov    0x1c(%rsi),%r14d
>>      573:    49 8b 8f 40 01 00 00     mov    0x140(%r15),%rcx
>>      57a:    48 89 f3                 mov    %rsi,%rbx
>>      57d:    45 85 f6                 test   %r14d,%r14d
>>      580:    0f 8e d5 00 00 00        jle    65b
>> <drm_calc_timestamping_constants+0x12b>
>>      586:    0f b7 43 2a              movzwl 0x2a(%rbx),%eax
>>      58a:    49 63 f6                 movslq %r14d,%rsi
>>      58d:    31 d2                    xor    %edx,%edx
>>      58f:    48 89 c7                 mov    %rax,%rdi
>>      592:    48 69 c0 40 42 0f 00     imul   $0xf4240,%rax,%rax
>>      599:    48 f7 f6                 div    %rsi
>>      59c:    31 d2                    xor    %edx,%edx
>>      59e:    48 89 45 d0              mov    %rax,-0x30(%rbp)
>>      5a2:    0f b7 43 38              movzwl 0x38(%rbx),%eax
>>      5a6:    0f af c7                 imul   %edi,%eax
>>      5a9:    48 98                    cltq
>>      5ab:    48 69 c0 40 42 0f 00     imul   $0xf4240,%rax,%rax
>>      5b2:    48 f7 f6                 div    %rsi
>>      5b5:    41 89 c5                 mov    %eax,%r13d
>>      5b8:    f6 43 18 10              testb  $0x10,0x18(%rbx)
>>      5bc:    74 0a                    je     5c8
>> <drm_calc_timestamping_constants+0x98>
>>      5be:    41 c1 ed 1f              shr    $0x1f,%r13d
>>      5c2:    41 01 c5                 add    %eax,%r13d
>>      5c5:    41 d1 fd                 sar    %r13d
>>      5c8:    4b 8d 04 c0              lea    (%r8,%r8,8),%rax
>>      5cc:    48 89 de                 mov    %rbx,%rsi
>>      5cf:    49 8d 3c 40              lea    (%r8,%rax,2),%rdi
>>      5d3:    8b 45 d0                 mov    -0x30(%rbp),%eax
>>      5d6:    48 c1 e7 04              shl    $0x4,%rdi
>>      5da:    48 01 cf                 add    %rcx,%rdi
>>      5dd:    89 47 78                 mov    %eax,0x78(%rdi)
>>      5e0:    48 83 ef 80              sub $0xffffffffffffff80,%rdi
>>      5e4:    44 89 6f f4              mov    %r13d,-0xc(%rdi)
>>      5e8:    e8 00 00 00 00           callq  5ed
>> <drm_calc_timestamping_constants+0xbd>
>>      5ed:    0f b7 53 2e              movzwl 0x2e(%rbx),%edx
>>      5f1:    0f b7 43 38              movzwl 0x38(%rbx),%eax
>>      5f5:    44 0f b7 4b 2a           movzwl 0x2a(%rbx),%r9d
>>      5fa:    45 8b 44 24 60           mov    0x60(%r12),%r8d
>>      5ff:    4d 85 ff                 test   %r15,%r15
>>      602:    0f 84 87 00 00 00        je     68f
>> <drm_calc_timestamping_constants+0x15f>
>>      608:    49 8b 77 08              mov    0x8(%r15),%rsi
>>      60c:    52                       push   %rdx
>>      60d:    31 ff                    xor    %edi,%edi
>>      60f:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>      616:    50                       push   %rax
>>      617:    31 d2                    xor    %edx,%edx
>>      619:    e8 00 00 00 00           callq  61e
>> <drm_calc_timestamping_constants+0xee>
>>      61e:    45 8b 44 24 60           mov    0x60(%r12),%r8d
>>      623:    4d 8b 7f 08              mov    0x8(%r15),%r15
>>      627:    5f                       pop    %rdi
>>      628:    41 59                    pop    %r9
>>      62a:    8b 45 d0                 mov    -0x30(%rbp),%eax
>>      62d:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>      634:    4c 89 fe                 mov    %r15,%rsi
>>      637:    45 89 f1                 mov    %r14d,%r9d
>>      63a:    31 d2                    xor    %edx,%edx
>>      63c:    31 ff                    xor    %edi,%edi
>>      63e:    50                       push   %rax
>>      63f:    41 55                    push   %r13
>>      641:    e8 00 00 00 00           callq  646
>> <drm_calc_timestamping_constants+0x116>
>>      646:    59                       pop    %rcx
>>      647:    5e                       pop    %rsi
>>      648:    48 8d 65 d8              lea    -0x28(%rbp),%rsp
>>      64c:    5b                       pop    %rbx
>>      64d:    41 5c                    pop    %r12
>>      64f:    41 5d                    pop    %r13
>>      651:    41 5e                    pop    %r14
>>      653:    41 5f                    pop    %r15
>>      655:    5d                       pop    %rbp
>>      656:    e9 00 00 00 00           jmpq   65b
>> <drm_calc_timestamping_constants+0x12b>
>>      65b:    41 8b 54 24 60           mov    0x60(%r12),%edx
>>      660:    49 8b 7f 08              mov    0x8(%r15),%rdi
>>      664:    44 89 45 c4              mov    %r8d,-0x3c(%rbp)
>>      668:    45 31 ed                 xor    %r13d,%r13d
>>      66b:    48 c7 c6 00 00 00 00     mov    $0x0,%rsi
>>      672:    48 89 4d c8              mov    %rcx,-0x38(%rbp)
>>      676:    e8 00 00 00 00           callq  67b
>> <drm_calc_timestamping_constants+0x14b>
>>      67b:    c7 45 d0 00 00 00 00     movl   $0x0,-0x30(%rbp)
>>      682:    44 8b 45 c4              mov    -0x3c(%rbp),%r8d
>>      686:    48 8b 4d c8              mov    -0x38(%rbp),%rcx
>>      68a:    e9 39 ff ff ff           jmpq   5c8
>> <drm_calc_timestamping_constants+0x98>
>>      68f:    52                       push   %rdx
>>      690:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>      697:    31 d2                    xor    %edx,%edx
>>      699:    31 f6                    xor    %esi,%esi
>>      69b:    50                       push   %rax
>>      69c:    31 ff                    xor    %edi,%edi
>>      69e:    e8 00 00 00 00           callq  6a3
>> <drm_calc_timestamping_constants+0x173>
>>      6a3:    45 8b 44 24 60           mov    0x60(%r12),%r8d
>>      6a8:    58                       pop    %rax
>>      6a9:    5a                       pop    %rdx
>>      6aa:    e9 7b ff ff ff           jmpq   62a
>> <drm_calc_timestamping_constants+0xfa>
>>      6af:    49 8b 7f 08              mov    0x8(%r15),%rdi
>>      6b3:    4c 8b 67 50              mov    0x50(%rdi),%r12
>>      6b7:    4d 85 e4                 test   %r12,%r12
>>      6ba:    74 25                    je     6e1
>> <drm_calc_timestamping_constants+0x1b1>
>>      6bc:    e8 00 00 00 00           callq  6c1
>> <drm_calc_timestamping_constants+0x191>
>>      6c1:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>      6c8:    4c 89 e2                 mov    %r12,%rdx
>>      6cb:    48 c7 c7 00 00 00 00     mov    $0x0,%rdi
>>      6d2:    48 89 c6                 mov    %rax,%rsi
>>      6d5:    e8 00 00 00 00           callq  6da
>> <drm_calc_timestamping_constants+0x1aa>
>>      6da:    0f 0b                    ud2
>>      6dc:    e9 67 ff ff ff           jmpq   648
>> <drm_calc_timestamping_constants+0x118>
>>      6e1:    4c 8b 27                 mov    (%rdi),%r12
>>      6e4:    eb d6                    jmp    6bc
>> <drm_calc_timestamping_constants+0x18c>
>>      6e6:    66 2e 0f 1f 84 00 00     nopw   %cs:0x0(%rax,%rax,1)
>>      6ed:    00 00 00
>>      6f0:    90                       nop
>>      6f1:    90                       nop
>>      6f2:    90                       nop
>>      6f3:    90                       nop
>>      6f4:    90                       nop
>>      6f5:    90                       nop
>>      6f6:    90                       nop
>>      6f7:    90                       nop
>>      6f8:    90                       nop
>>      6f9:    90                       nop
>>      6fa:    90                       nop
>>      6fb:    90                       nop
>>      6fc:    90                       nop
>>      6fd:    90                       nop
>>      6fe:    90                       nop
>>      6ff:    90                       nop
>>
>>
>>> David
>>>
>>> -
>>> Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
>>> Registration No: 1397386 (Wales)
>>>

2023-05-22 12:25:58

by Jani Nikula

[permalink] [raw]
Subject: Re: [PATCH] drm/drm_vblank.c: avoid unsigned int to signed int cast

On Mon, 22 May 2023, Sui Jingfeng <[email protected]> wrote:
> Hi,
>
> On 2023/5/22 19:29, Jani Nikula wrote:
>> On Thu, 18 May 2023, Sui Jingfeng <[email protected]> wrote:
>>> On 2023/5/17 18:59, David Laight wrote:
>>>> From: [email protected]
>>>>> Sent: 16 May 2023 18:30
>>>>>
>>>>> From: Sui Jingfeng <[email protected]>
>>>>>
>>>>> Both mode->crtc_htotal and mode->crtc_vtotal are u16 type,
>>>>> mode->crtc_htotal * mode->crtc_vtotal will results a unsigned type.
>>>> Nope, u16 gets promoted to 'signed int' and the result of the
>>>> multiply is also signed.
>>> I believe that signed or unsigned is dependent on the declaration.
>>>
>>> I am talk about the math, while you are talking about compiler.
>>>
>>> I admit that u16 gets promoted to 'signed int' is true, but this is
>>> irrelevant,
>>>
>>> the point is how to understand the returned value.
>>>
>>>
>>> How does the compiler generate the code is one thing, how do we
>>> interpret the result is another
>>>
>>> How does the compiler generate the code is NOT determined by us, while
>>> how do we interpret the result is determined by us.
>>>
>>>
>>> I believe that using a u32 type to interpret the result(u16 * u16) is
>>> always true, it is true in the perspective of *math*.
>>>
>>> Integer promotions is the details of C program language. If the result
>>> of the multiply is signed, then there are risks that
>>>
>>> the result is negative, what's the benefit to present this risk to the
>>> programmer?
>>>
>>> What's the benefit to tell me(and others) that u16 * u16 yield a signed
>>> value? and can be negative?
>>>
>>> Using int type as the return type bring concerns to the programmer and
>>> the user of the function,
>>>
>>> even though this is not impossible in practice.
>> In general, do not use unsigned types in arithmethic to avoid negative
>> values, because most people will be tripped over by integer promotion
>> rules, and you'll get negative values anyway.
>>
>> I'll bet most people will be surprised to see what this prints:
>>
>> #include <stdio.h>
>> #include <stdint.h>
>>
>> int main(void)
>> {
>> uint16_t x = 0xffff;
>> uint16_t y = 0xffff;
>> uint64_t z = x * y;
>>
>> printf("0x%016lx\n", z);
>> printf("%ld\n", z);
>
> Here, please replace the "%ld\n" with the "%lu\n", then you will see the
> difference.
>
> you are casting the variable 'z' to signed value,  "%d" is for printing
> signed value, and "%u" is for printing unsigned value.
>
>
> Your simple code explained exactly why you are still in confusion,

Am I?

Take a look at the values, and explain the math.


BR,
Jani.

>
> that is u16 * u16  can yield a negative value if you use the int as the
> return type. Because it overflowed.
>
>> printf("%d\n", x * y);
>> }
>>
>> And it's not that different from what you have below. Your patch doesn't
>> change anything, and doesn't make it any less confusing.
>>
>> BR,
>> Jani.
>>
>>
>>>>> Using a u32 is enough to store the result, but considering that the
>>>>> result will be casted to u64 soon after. We use a u64 type directly.
>>>>> So there no need to cast it to signed type and cast back then.
>>>> ....
>>>>> - int frame_size = mode->crtc_htotal * mode->crtc_vtotal;
>>>>> + u64 frame_size = mode->crtc_htotal * mode->crtc_vtotal;
>>>> ...
>>>>> - framedur_ns = div_u64((u64) frame_size * 1000000, dotclock);
>>>>> + framedur_ns = div_u64(frame_size * 1000000, dotclock);
>>>> The (u64) cast is there to extend the value to 64bits, not
>>>> because the original type is signed.
>>> Sorry about my expression, I think my sentence did not mention anything
>>> about 'because the original type is signed'.
>>>
>>> In the contrary, my patch eliminated the concerns to the reviewer. It
>>> say that the results of the multiply can't be negative.
>>>
>>> My intent is to tell the compiler we want a unsigned return type, but
>>> GCC emit 'imul' instruction for the multiply......
>>>
>>> I'm using u64 as the return type, because div_u64() function accept a
>>> u64 type value as its first argument.
>>>
>>>> The compiler will detect that the old code is a 32x32 multiply
>>>> where a 64bit result is needed, that may not be true for the
>>>> changed code (it would need to track back as far as the u16s).
>>> I don't believe my code could be wrong.
>>>
>>> when you use the word 'may', you are saying that it could be wrong after
>>> apply my patch.
>>>
>>> Then you have to find at least one test example to prove you point, in
>>> which case my codes generate wrong results.
>>>
>>> Again I don't believe you could find one.
>>>
>>>> It is not uncommon to force a 64bit result from a multiply
>>>> by making the constant 64bit. As in:
>>>> div_u64(frame_size * 1000000ULL, dotclock);
>>> In fact, After apply this patch, the ASM code generated is same with before.
>>>
>>> This may because the GCC is smart enough to generate optimized code in
>>> either case,
>>>
>>> I think It could be different with a different optimization-level.
>>>
>>> I have tested this patch on three different architecture,  I can not
>>> find error still.
>>>
>>> Below is the assembly extract on x86-64: because GCC generate the same
>>> code in either case,
>>>
>>> so I pasted only one copy here.
>>>
>>>
>>> 0000000000000530 <drm_calc_timestamping_constants>:
>>>      530:    f3 0f 1e fa              endbr64
>>>      534:    e8 00 00 00 00           callq  539
>>> <drm_calc_timestamping_constants+0x9>
>>>      539:    55                       push   %rbp
>>>      53a:    48 89 e5                 mov    %rsp,%rbp
>>>      53d:    41 57                    push   %r15
>>>      53f:    41 56                    push   %r14
>>>      541:    41 55                    push   %r13
>>>      543:    41 54                    push   %r12
>>>      545:    53                       push   %rbx
>>>      546:    48 83 ec 18              sub    $0x18,%rsp
>>>      54a:    4c 8b 3f                 mov    (%rdi),%r15
>>>      54d:    41 8b 87 6c 01 00 00     mov    0x16c(%r15),%eax
>>>      554:    85 c0                    test   %eax,%eax
>>>      556:    0f 84 ec 00 00 00        je     648
>>> <drm_calc_timestamping_constants+0x118>
>>>      55c:    44 8b 87 90 00 00 00     mov    0x90(%rdi),%r8d
>>>      563:    49 89 fc                 mov    %rdi,%r12
>>>      566:    44 39 c0                 cmp    %r8d,%eax
>>>      569:    0f 86 40 01 00 00        jbe    6af
>>> <drm_calc_timestamping_constants+0x17f>
>>>      56f:    44 8b 76 1c              mov    0x1c(%rsi),%r14d
>>>      573:    49 8b 8f 40 01 00 00     mov    0x140(%r15),%rcx
>>>      57a:    48 89 f3                 mov    %rsi,%rbx
>>>      57d:    45 85 f6                 test   %r14d,%r14d
>>>      580:    0f 8e d5 00 00 00        jle    65b
>>> <drm_calc_timestamping_constants+0x12b>
>>>      586:    0f b7 43 2a              movzwl 0x2a(%rbx),%eax
>>>      58a:    49 63 f6                 movslq %r14d,%rsi
>>>      58d:    31 d2                    xor    %edx,%edx
>>>      58f:    48 89 c7                 mov    %rax,%rdi
>>>      592:    48 69 c0 40 42 0f 00     imul   $0xf4240,%rax,%rax
>>>      599:    48 f7 f6                 div    %rsi
>>>      59c:    31 d2                    xor    %edx,%edx
>>>      59e:    48 89 45 d0              mov    %rax,-0x30(%rbp)
>>>      5a2:    0f b7 43 38              movzwl 0x38(%rbx),%eax
>>>      5a6:    0f af c7                 imul   %edi,%eax
>>>      5a9:    48 98                    cltq
>>>      5ab:    48 69 c0 40 42 0f 00     imul   $0xf4240,%rax,%rax
>>>      5b2:    48 f7 f6                 div    %rsi
>>>      5b5:    41 89 c5                 mov    %eax,%r13d
>>>      5b8:    f6 43 18 10              testb  $0x10,0x18(%rbx)
>>>      5bc:    74 0a                    je     5c8
>>> <drm_calc_timestamping_constants+0x98>
>>>      5be:    41 c1 ed 1f              shr    $0x1f,%r13d
>>>      5c2:    41 01 c5                 add    %eax,%r13d
>>>      5c5:    41 d1 fd                 sar    %r13d
>>>      5c8:    4b 8d 04 c0              lea    (%r8,%r8,8),%rax
>>>      5cc:    48 89 de                 mov    %rbx,%rsi
>>>      5cf:    49 8d 3c 40              lea    (%r8,%rax,2),%rdi
>>>      5d3:    8b 45 d0                 mov    -0x30(%rbp),%eax
>>>      5d6:    48 c1 e7 04              shl    $0x4,%rdi
>>>      5da:    48 01 cf                 add    %rcx,%rdi
>>>      5dd:    89 47 78                 mov    %eax,0x78(%rdi)
>>>      5e0:    48 83 ef 80              sub $0xffffffffffffff80,%rdi
>>>      5e4:    44 89 6f f4              mov    %r13d,-0xc(%rdi)
>>>      5e8:    e8 00 00 00 00           callq  5ed
>>> <drm_calc_timestamping_constants+0xbd>
>>>      5ed:    0f b7 53 2e              movzwl 0x2e(%rbx),%edx
>>>      5f1:    0f b7 43 38              movzwl 0x38(%rbx),%eax
>>>      5f5:    44 0f b7 4b 2a           movzwl 0x2a(%rbx),%r9d
>>>      5fa:    45 8b 44 24 60           mov    0x60(%r12),%r8d
>>>      5ff:    4d 85 ff                 test   %r15,%r15
>>>      602:    0f 84 87 00 00 00        je     68f
>>> <drm_calc_timestamping_constants+0x15f>
>>>      608:    49 8b 77 08              mov    0x8(%r15),%rsi
>>>      60c:    52                       push   %rdx
>>>      60d:    31 ff                    xor    %edi,%edi
>>>      60f:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>>      616:    50                       push   %rax
>>>      617:    31 d2                    xor    %edx,%edx
>>>      619:    e8 00 00 00 00           callq  61e
>>> <drm_calc_timestamping_constants+0xee>
>>>      61e:    45 8b 44 24 60           mov    0x60(%r12),%r8d
>>>      623:    4d 8b 7f 08              mov    0x8(%r15),%r15
>>>      627:    5f                       pop    %rdi
>>>      628:    41 59                    pop    %r9
>>>      62a:    8b 45 d0                 mov    -0x30(%rbp),%eax
>>>      62d:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>>      634:    4c 89 fe                 mov    %r15,%rsi
>>>      637:    45 89 f1                 mov    %r14d,%r9d
>>>      63a:    31 d2                    xor    %edx,%edx
>>>      63c:    31 ff                    xor    %edi,%edi
>>>      63e:    50                       push   %rax
>>>      63f:    41 55                    push   %r13
>>>      641:    e8 00 00 00 00           callq  646
>>> <drm_calc_timestamping_constants+0x116>
>>>      646:    59                       pop    %rcx
>>>      647:    5e                       pop    %rsi
>>>      648:    48 8d 65 d8              lea    -0x28(%rbp),%rsp
>>>      64c:    5b                       pop    %rbx
>>>      64d:    41 5c                    pop    %r12
>>>      64f:    41 5d                    pop    %r13
>>>      651:    41 5e                    pop    %r14
>>>      653:    41 5f                    pop    %r15
>>>      655:    5d                       pop    %rbp
>>>      656:    e9 00 00 00 00           jmpq   65b
>>> <drm_calc_timestamping_constants+0x12b>
>>>      65b:    41 8b 54 24 60           mov    0x60(%r12),%edx
>>>      660:    49 8b 7f 08              mov    0x8(%r15),%rdi
>>>      664:    44 89 45 c4              mov    %r8d,-0x3c(%rbp)
>>>      668:    45 31 ed                 xor    %r13d,%r13d
>>>      66b:    48 c7 c6 00 00 00 00     mov    $0x0,%rsi
>>>      672:    48 89 4d c8              mov    %rcx,-0x38(%rbp)
>>>      676:    e8 00 00 00 00           callq  67b
>>> <drm_calc_timestamping_constants+0x14b>
>>>      67b:    c7 45 d0 00 00 00 00     movl   $0x0,-0x30(%rbp)
>>>      682:    44 8b 45 c4              mov    -0x3c(%rbp),%r8d
>>>      686:    48 8b 4d c8              mov    -0x38(%rbp),%rcx
>>>      68a:    e9 39 ff ff ff           jmpq   5c8
>>> <drm_calc_timestamping_constants+0x98>
>>>      68f:    52                       push   %rdx
>>>      690:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>>      697:    31 d2                    xor    %edx,%edx
>>>      699:    31 f6                    xor    %esi,%esi
>>>      69b:    50                       push   %rax
>>>      69c:    31 ff                    xor    %edi,%edi
>>>      69e:    e8 00 00 00 00           callq  6a3
>>> <drm_calc_timestamping_constants+0x173>
>>>      6a3:    45 8b 44 24 60           mov    0x60(%r12),%r8d
>>>      6a8:    58                       pop    %rax
>>>      6a9:    5a                       pop    %rdx
>>>      6aa:    e9 7b ff ff ff           jmpq   62a
>>> <drm_calc_timestamping_constants+0xfa>
>>>      6af:    49 8b 7f 08              mov    0x8(%r15),%rdi
>>>      6b3:    4c 8b 67 50              mov    0x50(%rdi),%r12
>>>      6b7:    4d 85 e4                 test   %r12,%r12
>>>      6ba:    74 25                    je     6e1
>>> <drm_calc_timestamping_constants+0x1b1>
>>>      6bc:    e8 00 00 00 00           callq  6c1
>>> <drm_calc_timestamping_constants+0x191>
>>>      6c1:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>>      6c8:    4c 89 e2                 mov    %r12,%rdx
>>>      6cb:    48 c7 c7 00 00 00 00     mov    $0x0,%rdi
>>>      6d2:    48 89 c6                 mov    %rax,%rsi
>>>      6d5:    e8 00 00 00 00           callq  6da
>>> <drm_calc_timestamping_constants+0x1aa>
>>>      6da:    0f 0b                    ud2
>>>      6dc:    e9 67 ff ff ff           jmpq   648
>>> <drm_calc_timestamping_constants+0x118>
>>>      6e1:    4c 8b 27                 mov    (%rdi),%r12
>>>      6e4:    eb d6                    jmp    6bc
>>> <drm_calc_timestamping_constants+0x18c>
>>>      6e6:    66 2e 0f 1f 84 00 00     nopw   %cs:0x0(%rax,%rax,1)
>>>      6ed:    00 00 00
>>>      6f0:    90                       nop
>>>      6f1:    90                       nop
>>>      6f2:    90                       nop
>>>      6f3:    90                       nop
>>>      6f4:    90                       nop
>>>      6f5:    90                       nop
>>>      6f6:    90                       nop
>>>      6f7:    90                       nop
>>>      6f8:    90                       nop
>>>      6f9:    90                       nop
>>>      6fa:    90                       nop
>>>      6fb:    90                       nop
>>>      6fc:    90                       nop
>>>      6fd:    90                       nop
>>>      6fe:    90                       nop
>>>      6ff:    90                       nop
>>>
>>>
>>>> David
>>>>
>>>> -
>>>> Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
>>>> Registration No: 1397386 (Wales)
>>>>

--
Jani Nikula, Intel Open Source Graphics Center

2023-05-22 12:27:16

by David Laight

[permalink] [raw]
Subject: RE: [PATCH] drm/drm_vblank.c: avoid unsigned int to signed int cast

From: [email protected] <[email protected]>
> Sent: 22 May 2023 12:56
...
> > I'll bet most people will be surprised to see what this prints:
> >
> > #include <stdio.h>
> > #include <stdint.h>
> >
> > int main(void)
> > {
> > uint16_t x = 0xffff;
> > uint16_t y = 0xffff;
> > uint64_t z = x * y;
> >
> > printf("0x%016lx\n", z);
> > printf("%ld\n", z);
>
> Here, please replace the "%ld\n" with the "%lu\n", then you will see the
> difference.
>
> you are casting the variable 'z' to signed value,  "%d" is for printing
> signed value, and "%u" is for printing unsigned value.

That makes very little difference on 2's compliment systems.
They both display the contents of the variable.

> Your simple code explained exactly why you are still in confusion,
>
> that is u16 * u16  can yield a negative value if you use the int as the
> return type. Because it overflowed.

There is no 'return type', the type of 'u16 * u16' is signed int.
When 'signed int' is promoted/cast to u64 it is first sign extended
to 64 bits.

You can get what you want/expect by either forcing an unsigned multiply
or by explicitly casting the result of the multiply to u32.
So the product in 'z = (x + 0u) * y' is 'unsigned int' it gets
promoted to int64_t (ie a signed type) and then converted to
unsigned.

David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)

2023-05-22 13:05:43

by Sui Jingfeng

[permalink] [raw]
Subject: Re: [PATCH] drm/drm_vblank.c: avoid unsigned int to signed int cast

Hi,

On 2023/5/22 20:13, Jani Nikula wrote:
> On Mon, 22 May 2023, Sui Jingfeng <[email protected]> wrote:
>> Hi,
>>
>> On 2023/5/22 19:29, Jani Nikula wrote:
>>> On Thu, 18 May 2023, Sui Jingfeng <[email protected]> wrote:
>>>> On 2023/5/17 18:59, David Laight wrote:
>>>>> From: [email protected]
>>>>>> Sent: 16 May 2023 18:30
>>>>>>
>>>>>> From: Sui Jingfeng <[email protected]>
>>>>>>
>>>>>> Both mode->crtc_htotal and mode->crtc_vtotal are u16 type,
>>>>>> mode->crtc_htotal * mode->crtc_vtotal will results a unsigned type.
>>>>> Nope, u16 gets promoted to 'signed int' and the result of the
>>>>> multiply is also signed.
>>>> I believe that signed or unsigned is dependent on the declaration.
>>>>
>>>> I am talk about the math, while you are talking about compiler.
>>>>
>>>> I admit that u16 gets promoted to 'signed int' is true, but this is
>>>> irrelevant,
>>>>
>>>> the point is how to understand the returned value.
>>>>
>>>>
>>>> How does the compiler generate the code is one thing, how do we
>>>> interpret the result is another
>>>>
>>>> How does the compiler generate the code is NOT determined by us, while
>>>> how do we interpret the result is determined by us.
>>>>
>>>>
>>>> I believe that using a u32 type to interpret the result(u16 * u16) is
>>>> always true, it is true in the perspective of *math*.
>>>>
>>>> Integer promotions is the details of C program language. If the result
>>>> of the multiply is signed, then there are risks that
>>>>
>>>> the result is negative, what's the benefit to present this risk to the
>>>> programmer?
>>>>
>>>> What's the benefit to tell me(and others) that u16 * u16 yield a signed
>>>> value? and can be negative?
>>>>
>>>> Using int type as the return type bring concerns to the programmer and
>>>> the user of the function,
>>>>
>>>> even though this is not impossible in practice.
>>> In general, do not use unsigned types in arithmethic to avoid negative
>>> values, because most people will be tripped over by integer promotion
>>> rules, and you'll get negative values anyway.
>>>
>>> I'll bet most people will be surprised to see what this prints:
>>>
>>> #include <stdio.h>
>>> #include <stdint.h>
>>>
>>> int main(void)
>>> {
>>> uint16_t x = 0xffff;
>>> uint16_t y = 0xffff;
>>> uint64_t z = x * y;
>>>
>>> printf("0x%016lx\n", z);
>>> printf("%ld\n", z);
>> Here, please replace the "%ld\n" with the "%lu\n", then you will see the
>> difference.
>>
>> you are casting the variable 'z' to signed value,  "%d" is for printing
>> signed value, and "%u" is for printing unsigned value.
>>
>>
>> Your simple code explained exactly why you are still in confusion,
> Am I?
>
> Take a look at the values, and explain the math.

I meant the value itself is represent with 2's compliment,

when you print a value with '%ld', then you will get the signed version,

when you print a value with '%lu', then you will get the unsigned version.

The result of a u16*u16 couldn't be negative in math.


But when you using a '%ld' or '%d' to print a unsigned value, then is wrong.

This is also the case which you shouldn't using a int type to store the result of u16*u16.

because when I seen a int type, I will choose '%d' to print it,

when I seen a unsigned int type, I will choose '%u' to print it.

when using a int type as the return type, this could lead people to using '%d' to print

such a value. Then, it generate the confusion as this little test program shows.

>
> BR,
> Jani.
>
>> that is u16 * u16  can yield a negative value if you use the int as the
>> return type. Because it overflowed.
>>
>>> printf("%d\n", x * y);
>>> }
>>>
>>> And it's not that different from what you have below. Your patch doesn't
>>> change anything, and doesn't make it any less confusing.
>>>
>>> BR,
>>> Jani.
>>>
>>>
>>>>>> Using a u32 is enough to store the result, but considering that the
>>>>>> result will be casted to u64 soon after. We use a u64 type directly.
>>>>>> So there no need to cast it to signed type and cast back then.
>>>>> ....
>>>>>> - int frame_size = mode->crtc_htotal * mode->crtc_vtotal;
>>>>>> + u64 frame_size = mode->crtc_htotal * mode->crtc_vtotal;
>>>>> ...
>>>>>> - framedur_ns = div_u64((u64) frame_size * 1000000, dotclock);
>>>>>> + framedur_ns = div_u64(frame_size * 1000000, dotclock);
>>>>> The (u64) cast is there to extend the value to 64bits, not
>>>>> because the original type is signed.
>>>> Sorry about my expression, I think my sentence did not mention anything
>>>> about 'because the original type is signed'.
>>>>
>>>> In the contrary, my patch eliminated the concerns to the reviewer. It
>>>> say that the results of the multiply can't be negative.
>>>>
>>>> My intent is to tell the compiler we want a unsigned return type, but
>>>> GCC emit 'imul' instruction for the multiply......
>>>>
>>>> I'm using u64 as the return type, because div_u64() function accept a
>>>> u64 type value as its first argument.
>>>>
>>>>> The compiler will detect that the old code is a 32x32 multiply
>>>>> where a 64bit result is needed, that may not be true for the
>>>>> changed code (it would need to track back as far as the u16s).
>>>> I don't believe my code could be wrong.
>>>>
>>>> when you use the word 'may', you are saying that it could be wrong after
>>>> apply my patch.
>>>>
>>>> Then you have to find at least one test example to prove you point, in
>>>> which case my codes generate wrong results.
>>>>
>>>> Again I don't believe you could find one.
>>>>
>>>>> It is not uncommon to force a 64bit result from a multiply
>>>>> by making the constant 64bit. As in:
>>>>> div_u64(frame_size * 1000000ULL, dotclock);
>>>> In fact, After apply this patch, the ASM code generated is same with before.
>>>>
>>>> This may because the GCC is smart enough to generate optimized code in
>>>> either case,
>>>>
>>>> I think It could be different with a different optimization-level.
>>>>
>>>> I have tested this patch on three different architecture,  I can not
>>>> find error still.
>>>>
>>>> Below is the assembly extract on x86-64: because GCC generate the same
>>>> code in either case,
>>>>
>>>> so I pasted only one copy here.
>>>>
>>>>
>>>> 0000000000000530 <drm_calc_timestamping_constants>:
>>>>      530:    f3 0f 1e fa              endbr64
>>>>      534:    e8 00 00 00 00           callq  539
>>>> <drm_calc_timestamping_constants+0x9>
>>>>      539:    55                       push   %rbp
>>>>      53a:    48 89 e5                 mov    %rsp,%rbp
>>>>      53d:    41 57                    push   %r15
>>>>      53f:    41 56                    push   %r14
>>>>      541:    41 55                    push   %r13
>>>>      543:    41 54                    push   %r12
>>>>      545:    53                       push   %rbx
>>>>      546:    48 83 ec 18              sub    $0x18,%rsp
>>>>      54a:    4c 8b 3f                 mov    (%rdi),%r15
>>>>      54d:    41 8b 87 6c 01 00 00     mov    0x16c(%r15),%eax
>>>>      554:    85 c0                    test   %eax,%eax
>>>>      556:    0f 84 ec 00 00 00        je     648
>>>> <drm_calc_timestamping_constants+0x118>
>>>>      55c:    44 8b 87 90 00 00 00     mov    0x90(%rdi),%r8d
>>>>      563:    49 89 fc                 mov    %rdi,%r12
>>>>      566:    44 39 c0                 cmp    %r8d,%eax
>>>>      569:    0f 86 40 01 00 00        jbe    6af
>>>> <drm_calc_timestamping_constants+0x17f>
>>>>      56f:    44 8b 76 1c              mov    0x1c(%rsi),%r14d
>>>>      573:    49 8b 8f 40 01 00 00     mov    0x140(%r15),%rcx
>>>>      57a:    48 89 f3                 mov    %rsi,%rbx
>>>>      57d:    45 85 f6                 test   %r14d,%r14d
>>>>      580:    0f 8e d5 00 00 00        jle    65b
>>>> <drm_calc_timestamping_constants+0x12b>
>>>>      586:    0f b7 43 2a              movzwl 0x2a(%rbx),%eax
>>>>      58a:    49 63 f6                 movslq %r14d,%rsi
>>>>      58d:    31 d2                    xor    %edx,%edx
>>>>      58f:    48 89 c7                 mov    %rax,%rdi
>>>>      592:    48 69 c0 40 42 0f 00     imul   $0xf4240,%rax,%rax
>>>>      599:    48 f7 f6                 div    %rsi
>>>>      59c:    31 d2                    xor    %edx,%edx
>>>>      59e:    48 89 45 d0              mov    %rax,-0x30(%rbp)
>>>>      5a2:    0f b7 43 38              movzwl 0x38(%rbx),%eax
>>>>      5a6:    0f af c7                 imul   %edi,%eax
>>>>      5a9:    48 98                    cltq
>>>>      5ab:    48 69 c0 40 42 0f 00     imul   $0xf4240,%rax,%rax
>>>>      5b2:    48 f7 f6                 div    %rsi
>>>>      5b5:    41 89 c5                 mov    %eax,%r13d
>>>>      5b8:    f6 43 18 10              testb  $0x10,0x18(%rbx)
>>>>      5bc:    74 0a                    je     5c8
>>>> <drm_calc_timestamping_constants+0x98>
>>>>      5be:    41 c1 ed 1f              shr    $0x1f,%r13d
>>>>      5c2:    41 01 c5                 add    %eax,%r13d
>>>>      5c5:    41 d1 fd                 sar    %r13d
>>>>      5c8:    4b 8d 04 c0              lea    (%r8,%r8,8),%rax
>>>>      5cc:    48 89 de                 mov    %rbx,%rsi
>>>>      5cf:    49 8d 3c 40              lea    (%r8,%rax,2),%rdi
>>>>      5d3:    8b 45 d0                 mov    -0x30(%rbp),%eax
>>>>      5d6:    48 c1 e7 04              shl    $0x4,%rdi
>>>>      5da:    48 01 cf                 add    %rcx,%rdi
>>>>      5dd:    89 47 78                 mov    %eax,0x78(%rdi)
>>>>      5e0:    48 83 ef 80              sub $0xffffffffffffff80,%rdi
>>>>      5e4:    44 89 6f f4              mov    %r13d,-0xc(%rdi)
>>>>      5e8:    e8 00 00 00 00           callq  5ed
>>>> <drm_calc_timestamping_constants+0xbd>
>>>>      5ed:    0f b7 53 2e              movzwl 0x2e(%rbx),%edx
>>>>      5f1:    0f b7 43 38              movzwl 0x38(%rbx),%eax
>>>>      5f5:    44 0f b7 4b 2a           movzwl 0x2a(%rbx),%r9d
>>>>      5fa:    45 8b 44 24 60           mov    0x60(%r12),%r8d
>>>>      5ff:    4d 85 ff                 test   %r15,%r15
>>>>      602:    0f 84 87 00 00 00        je     68f
>>>> <drm_calc_timestamping_constants+0x15f>
>>>>      608:    49 8b 77 08              mov    0x8(%r15),%rsi
>>>>      60c:    52                       push   %rdx
>>>>      60d:    31 ff                    xor    %edi,%edi
>>>>      60f:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>>>      616:    50                       push   %rax
>>>>      617:    31 d2                    xor    %edx,%edx
>>>>      619:    e8 00 00 00 00           callq  61e
>>>> <drm_calc_timestamping_constants+0xee>
>>>>      61e:    45 8b 44 24 60           mov    0x60(%r12),%r8d
>>>>      623:    4d 8b 7f 08              mov    0x8(%r15),%r15
>>>>      627:    5f                       pop    %rdi
>>>>      628:    41 59                    pop    %r9
>>>>      62a:    8b 45 d0                 mov    -0x30(%rbp),%eax
>>>>      62d:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>>>      634:    4c 89 fe                 mov    %r15,%rsi
>>>>      637:    45 89 f1                 mov    %r14d,%r9d
>>>>      63a:    31 d2                    xor    %edx,%edx
>>>>      63c:    31 ff                    xor    %edi,%edi
>>>>      63e:    50                       push   %rax
>>>>      63f:    41 55                    push   %r13
>>>>      641:    e8 00 00 00 00           callq  646
>>>> <drm_calc_timestamping_constants+0x116>
>>>>      646:    59                       pop    %rcx
>>>>      647:    5e                       pop    %rsi
>>>>      648:    48 8d 65 d8              lea    -0x28(%rbp),%rsp
>>>>      64c:    5b                       pop    %rbx
>>>>      64d:    41 5c                    pop    %r12
>>>>      64f:    41 5d                    pop    %r13
>>>>      651:    41 5e                    pop    %r14
>>>>      653:    41 5f                    pop    %r15
>>>>      655:    5d                       pop    %rbp
>>>>      656:    e9 00 00 00 00           jmpq   65b
>>>> <drm_calc_timestamping_constants+0x12b>
>>>>      65b:    41 8b 54 24 60           mov    0x60(%r12),%edx
>>>>      660:    49 8b 7f 08              mov    0x8(%r15),%rdi
>>>>      664:    44 89 45 c4              mov    %r8d,-0x3c(%rbp)
>>>>      668:    45 31 ed                 xor    %r13d,%r13d
>>>>      66b:    48 c7 c6 00 00 00 00     mov    $0x0,%rsi
>>>>      672:    48 89 4d c8              mov    %rcx,-0x38(%rbp)
>>>>      676:    e8 00 00 00 00           callq  67b
>>>> <drm_calc_timestamping_constants+0x14b>
>>>>      67b:    c7 45 d0 00 00 00 00     movl   $0x0,-0x30(%rbp)
>>>>      682:    44 8b 45 c4              mov    -0x3c(%rbp),%r8d
>>>>      686:    48 8b 4d c8              mov    -0x38(%rbp),%rcx
>>>>      68a:    e9 39 ff ff ff           jmpq   5c8
>>>> <drm_calc_timestamping_constants+0x98>
>>>>      68f:    52                       push   %rdx
>>>>      690:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>>>      697:    31 d2                    xor    %edx,%edx
>>>>      699:    31 f6                    xor    %esi,%esi
>>>>      69b:    50                       push   %rax
>>>>      69c:    31 ff                    xor    %edi,%edi
>>>>      69e:    e8 00 00 00 00           callq  6a3
>>>> <drm_calc_timestamping_constants+0x173>
>>>>      6a3:    45 8b 44 24 60           mov    0x60(%r12),%r8d
>>>>      6a8:    58                       pop    %rax
>>>>      6a9:    5a                       pop    %rdx
>>>>      6aa:    e9 7b ff ff ff           jmpq   62a
>>>> <drm_calc_timestamping_constants+0xfa>
>>>>      6af:    49 8b 7f 08              mov    0x8(%r15),%rdi
>>>>      6b3:    4c 8b 67 50              mov    0x50(%rdi),%r12
>>>>      6b7:    4d 85 e4                 test   %r12,%r12
>>>>      6ba:    74 25                    je     6e1
>>>> <drm_calc_timestamping_constants+0x1b1>
>>>>      6bc:    e8 00 00 00 00           callq  6c1
>>>> <drm_calc_timestamping_constants+0x191>
>>>>      6c1:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>>>      6c8:    4c 89 e2                 mov    %r12,%rdx
>>>>      6cb:    48 c7 c7 00 00 00 00     mov    $0x0,%rdi
>>>>      6d2:    48 89 c6                 mov    %rax,%rsi
>>>>      6d5:    e8 00 00 00 00           callq  6da
>>>> <drm_calc_timestamping_constants+0x1aa>
>>>>      6da:    0f 0b                    ud2
>>>>      6dc:    e9 67 ff ff ff           jmpq   648
>>>> <drm_calc_timestamping_constants+0x118>
>>>>      6e1:    4c 8b 27                 mov    (%rdi),%r12
>>>>      6e4:    eb d6                    jmp    6bc
>>>> <drm_calc_timestamping_constants+0x18c>
>>>>      6e6:    66 2e 0f 1f 84 00 00     nopw   %cs:0x0(%rax,%rax,1)
>>>>      6ed:    00 00 00
>>>>      6f0:    90                       nop
>>>>      6f1:    90                       nop
>>>>      6f2:    90                       nop
>>>>      6f3:    90                       nop
>>>>      6f4:    90                       nop
>>>>      6f5:    90                       nop
>>>>      6f6:    90                       nop
>>>>      6f7:    90                       nop
>>>>      6f8:    90                       nop
>>>>      6f9:    90                       nop
>>>>      6fa:    90                       nop
>>>>      6fb:    90                       nop
>>>>      6fc:    90                       nop
>>>>      6fd:    90                       nop
>>>>      6fe:    90                       nop
>>>>      6ff:    90                       nop
>>>>
>>>>
>>>>> David
>>>>>
>>>>> -
>>>>> Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
>>>>> Registration No: 1397386 (Wales)
>>>>>

2023-05-22 15:13:46

by Jani Nikula

[permalink] [raw]
Subject: Re: [PATCH] drm/drm_vblank.c: avoid unsigned int to signed int cast

On Mon, 22 May 2023, Sui Jingfeng <[email protected]> wrote:
> Hi,
>
> On 2023/5/22 20:13, Jani Nikula wrote:
>> On Mon, 22 May 2023, Sui Jingfeng <[email protected]> wrote:
>>> Hi,
>>>
>>> On 2023/5/22 19:29, Jani Nikula wrote:
>>>> On Thu, 18 May 2023, Sui Jingfeng <[email protected]> wrote:
>>>>> On 2023/5/17 18:59, David Laight wrote:
>>>>>> From: [email protected]
>>>>>>> Sent: 16 May 2023 18:30
>>>>>>>
>>>>>>> From: Sui Jingfeng <[email protected]>
>>>>>>>
>>>>>>> Both mode->crtc_htotal and mode->crtc_vtotal are u16 type,
>>>>>>> mode->crtc_htotal * mode->crtc_vtotal will results a unsigned type.
>>>>>> Nope, u16 gets promoted to 'signed int' and the result of the
>>>>>> multiply is also signed.
>>>>> I believe that signed or unsigned is dependent on the declaration.
>>>>>
>>>>> I am talk about the math, while you are talking about compiler.
>>>>>
>>>>> I admit that u16 gets promoted to 'signed int' is true, but this is
>>>>> irrelevant,
>>>>>
>>>>> the point is how to understand the returned value.
>>>>>
>>>>>
>>>>> How does the compiler generate the code is one thing, how do we
>>>>> interpret the result is another
>>>>>
>>>>> How does the compiler generate the code is NOT determined by us, while
>>>>> how do we interpret the result is determined by us.
>>>>>
>>>>>
>>>>> I believe that using a u32 type to interpret the result(u16 * u16) is
>>>>> always true, it is true in the perspective of *math*.
>>>>>
>>>>> Integer promotions is the details of C program language. If the result
>>>>> of the multiply is signed, then there are risks that
>>>>>
>>>>> the result is negative, what's the benefit to present this risk to the
>>>>> programmer?
>>>>>
>>>>> What's the benefit to tell me(and others) that u16 * u16 yield a signed
>>>>> value? and can be negative?
>>>>>
>>>>> Using int type as the return type bring concerns to the programmer and
>>>>> the user of the function,
>>>>>
>>>>> even though this is not impossible in practice.
>>>> In general, do not use unsigned types in arithmethic to avoid negative
>>>> values, because most people will be tripped over by integer promotion
>>>> rules, and you'll get negative values anyway.
>>>>
>>>> I'll bet most people will be surprised to see what this prints:
>>>>
>>>> #include <stdio.h>
>>>> #include <stdint.h>
>>>>
>>>> int main(void)
>>>> {
>>>> uint16_t x = 0xffff;
>>>> uint16_t y = 0xffff;
>>>> uint64_t z = x * y;
>>>>
>>>> printf("0x%016lx\n", z);
>>>> printf("%ld\n", z);
>>> Here, please replace the "%ld\n" with the "%lu\n", then you will see the
>>> difference.
>>>
>>> you are casting the variable 'z' to signed value,  "%d" is for printing
>>> signed value, and "%u" is for printing unsigned value.
>>>
>>>
>>> Your simple code explained exactly why you are still in confusion,
>> Am I?
>>
>> Take a look at the values, and explain the math.
>
> I meant the value itself is represent with 2's compliment,
>
> when you print a value with '%ld', then you will get the signed version,
>
> when you print a value with '%lu', then you will get the unsigned version.
>
> The result of a u16*u16 couldn't be negative in math.
>
>
> But when you using a '%ld' or '%d' to print a unsigned value, then is wrong.
>
> This is also the case which you shouldn't using a int type to store the result of u16*u16.
>
> because when I seen a int type, I will choose '%d' to print it,
>
> when I seen a unsigned int type, I will choose '%u' to print it.
>
> when using a int type as the return type, this could lead people to using '%d' to print
>
> such a value. Then, it generate the confusion as this little test program shows.

Using 0x%016lx and %lu results in 0xfffffffffffe0001 and
18446744073709420545, respectively. They are equal. They are indeed not
negative.

However 0xffff * 0xffff = 0xfffe0001. Or 4294836225 in decimal.

No matter what the math says, this is what actually happens in C.

I don't know what more I could possibly tell you.


BR,
Jani.


>
>>
>> BR,
>> Jani.
>>
>>> that is u16 * u16  can yield a negative value if you use the int as the
>>> return type. Because it overflowed.
>>>
>>>> printf("%d\n", x * y);
>>>> }
>>>>
>>>> And it's not that different from what you have below. Your patch doesn't
>>>> change anything, and doesn't make it any less confusing.
>>>>
>>>> BR,
>>>> Jani.
>>>>
>>>>
>>>>>>> Using a u32 is enough to store the result, but considering that the
>>>>>>> result will be casted to u64 soon after. We use a u64 type directly.
>>>>>>> So there no need to cast it to signed type and cast back then.
>>>>>> ....
>>>>>>> - int frame_size = mode->crtc_htotal * mode->crtc_vtotal;
>>>>>>> + u64 frame_size = mode->crtc_htotal * mode->crtc_vtotal;
>>>>>> ...
>>>>>>> - framedur_ns = div_u64((u64) frame_size * 1000000, dotclock);
>>>>>>> + framedur_ns = div_u64(frame_size * 1000000, dotclock);
>>>>>> The (u64) cast is there to extend the value to 64bits, not
>>>>>> because the original type is signed.
>>>>> Sorry about my expression, I think my sentence did not mention anything
>>>>> about 'because the original type is signed'.
>>>>>
>>>>> In the contrary, my patch eliminated the concerns to the reviewer. It
>>>>> say that the results of the multiply can't be negative.
>>>>>
>>>>> My intent is to tell the compiler we want a unsigned return type, but
>>>>> GCC emit 'imul' instruction for the multiply......
>>>>>
>>>>> I'm using u64 as the return type, because div_u64() function accept a
>>>>> u64 type value as its first argument.
>>>>>
>>>>>> The compiler will detect that the old code is a 32x32 multiply
>>>>>> where a 64bit result is needed, that may not be true for the
>>>>>> changed code (it would need to track back as far as the u16s).
>>>>> I don't believe my code could be wrong.
>>>>>
>>>>> when you use the word 'may', you are saying that it could be wrong after
>>>>> apply my patch.
>>>>>
>>>>> Then you have to find at least one test example to prove you point, in
>>>>> which case my codes generate wrong results.
>>>>>
>>>>> Again I don't believe you could find one.
>>>>>
>>>>>> It is not uncommon to force a 64bit result from a multiply
>>>>>> by making the constant 64bit. As in:
>>>>>> div_u64(frame_size * 1000000ULL, dotclock);
>>>>> In fact, After apply this patch, the ASM code generated is same with before.
>>>>>
>>>>> This may because the GCC is smart enough to generate optimized code in
>>>>> either case,
>>>>>
>>>>> I think It could be different with a different optimization-level.
>>>>>
>>>>> I have tested this patch on three different architecture,  I can not
>>>>> find error still.
>>>>>
>>>>> Below is the assembly extract on x86-64: because GCC generate the same
>>>>> code in either case,
>>>>>
>>>>> so I pasted only one copy here.
>>>>>
>>>>>
>>>>> 0000000000000530 <drm_calc_timestamping_constants>:
>>>>>      530:    f3 0f 1e fa              endbr64
>>>>>      534:    e8 00 00 00 00           callq  539
>>>>> <drm_calc_timestamping_constants+0x9>
>>>>>      539:    55                       push   %rbp
>>>>>      53a:    48 89 e5                 mov    %rsp,%rbp
>>>>>      53d:    41 57                    push   %r15
>>>>>      53f:    41 56                    push   %r14
>>>>>      541:    41 55                    push   %r13
>>>>>      543:    41 54                    push   %r12
>>>>>      545:    53                       push   %rbx
>>>>>      546:    48 83 ec 18              sub    $0x18,%rsp
>>>>>      54a:    4c 8b 3f                 mov    (%rdi),%r15
>>>>>      54d:    41 8b 87 6c 01 00 00     mov    0x16c(%r15),%eax
>>>>>      554:    85 c0                    test   %eax,%eax
>>>>>      556:    0f 84 ec 00 00 00        je     648
>>>>> <drm_calc_timestamping_constants+0x118>
>>>>>      55c:    44 8b 87 90 00 00 00     mov    0x90(%rdi),%r8d
>>>>>      563:    49 89 fc                 mov    %rdi,%r12
>>>>>      566:    44 39 c0                 cmp    %r8d,%eax
>>>>>      569:    0f 86 40 01 00 00        jbe    6af
>>>>> <drm_calc_timestamping_constants+0x17f>
>>>>>      56f:    44 8b 76 1c              mov    0x1c(%rsi),%r14d
>>>>>      573:    49 8b 8f 40 01 00 00     mov    0x140(%r15),%rcx
>>>>>      57a:    48 89 f3                 mov    %rsi,%rbx
>>>>>      57d:    45 85 f6                 test   %r14d,%r14d
>>>>>      580:    0f 8e d5 00 00 00        jle    65b
>>>>> <drm_calc_timestamping_constants+0x12b>
>>>>>      586:    0f b7 43 2a              movzwl 0x2a(%rbx),%eax
>>>>>      58a:    49 63 f6                 movslq %r14d,%rsi
>>>>>      58d:    31 d2                    xor    %edx,%edx
>>>>>      58f:    48 89 c7                 mov    %rax,%rdi
>>>>>      592:    48 69 c0 40 42 0f 00     imul   $0xf4240,%rax,%rax
>>>>>      599:    48 f7 f6                 div    %rsi
>>>>>      59c:    31 d2                    xor    %edx,%edx
>>>>>      59e:    48 89 45 d0              mov    %rax,-0x30(%rbp)
>>>>>      5a2:    0f b7 43 38              movzwl 0x38(%rbx),%eax
>>>>>      5a6:    0f af c7                 imul   %edi,%eax
>>>>>      5a9:    48 98                    cltq
>>>>>      5ab:    48 69 c0 40 42 0f 00     imul   $0xf4240,%rax,%rax
>>>>>      5b2:    48 f7 f6                 div    %rsi
>>>>>      5b5:    41 89 c5                 mov    %eax,%r13d
>>>>>      5b8:    f6 43 18 10              testb  $0x10,0x18(%rbx)
>>>>>      5bc:    74 0a                    je     5c8
>>>>> <drm_calc_timestamping_constants+0x98>
>>>>>      5be:    41 c1 ed 1f              shr    $0x1f,%r13d
>>>>>      5c2:    41 01 c5                 add    %eax,%r13d
>>>>>      5c5:    41 d1 fd                 sar    %r13d
>>>>>      5c8:    4b 8d 04 c0              lea    (%r8,%r8,8),%rax
>>>>>      5cc:    48 89 de                 mov    %rbx,%rsi
>>>>>      5cf:    49 8d 3c 40              lea    (%r8,%rax,2),%rdi
>>>>>      5d3:    8b 45 d0                 mov    -0x30(%rbp),%eax
>>>>>      5d6:    48 c1 e7 04              shl    $0x4,%rdi
>>>>>      5da:    48 01 cf                 add    %rcx,%rdi
>>>>>      5dd:    89 47 78                 mov    %eax,0x78(%rdi)
>>>>>      5e0:    48 83 ef 80              sub $0xffffffffffffff80,%rdi
>>>>>      5e4:    44 89 6f f4              mov    %r13d,-0xc(%rdi)
>>>>>      5e8:    e8 00 00 00 00           callq  5ed
>>>>> <drm_calc_timestamping_constants+0xbd>
>>>>>      5ed:    0f b7 53 2e              movzwl 0x2e(%rbx),%edx
>>>>>      5f1:    0f b7 43 38              movzwl 0x38(%rbx),%eax
>>>>>      5f5:    44 0f b7 4b 2a           movzwl 0x2a(%rbx),%r9d
>>>>>      5fa:    45 8b 44 24 60           mov    0x60(%r12),%r8d
>>>>>      5ff:    4d 85 ff                 test   %r15,%r15
>>>>>      602:    0f 84 87 00 00 00        je     68f
>>>>> <drm_calc_timestamping_constants+0x15f>
>>>>>      608:    49 8b 77 08              mov    0x8(%r15),%rsi
>>>>>      60c:    52                       push   %rdx
>>>>>      60d:    31 ff                    xor    %edi,%edi
>>>>>      60f:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>>>>      616:    50                       push   %rax
>>>>>      617:    31 d2                    xor    %edx,%edx
>>>>>      619:    e8 00 00 00 00           callq  61e
>>>>> <drm_calc_timestamping_constants+0xee>
>>>>>      61e:    45 8b 44 24 60           mov    0x60(%r12),%r8d
>>>>>      623:    4d 8b 7f 08              mov    0x8(%r15),%r15
>>>>>      627:    5f                       pop    %rdi
>>>>>      628:    41 59                    pop    %r9
>>>>>      62a:    8b 45 d0                 mov    -0x30(%rbp),%eax
>>>>>      62d:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>>>>      634:    4c 89 fe                 mov    %r15,%rsi
>>>>>      637:    45 89 f1                 mov    %r14d,%r9d
>>>>>      63a:    31 d2                    xor    %edx,%edx
>>>>>      63c:    31 ff                    xor    %edi,%edi
>>>>>      63e:    50                       push   %rax
>>>>>      63f:    41 55                    push   %r13
>>>>>      641:    e8 00 00 00 00           callq  646
>>>>> <drm_calc_timestamping_constants+0x116>
>>>>>      646:    59                       pop    %rcx
>>>>>      647:    5e                       pop    %rsi
>>>>>      648:    48 8d 65 d8              lea    -0x28(%rbp),%rsp
>>>>>      64c:    5b                       pop    %rbx
>>>>>      64d:    41 5c                    pop    %r12
>>>>>      64f:    41 5d                    pop    %r13
>>>>>      651:    41 5e                    pop    %r14
>>>>>      653:    41 5f                    pop    %r15
>>>>>      655:    5d                       pop    %rbp
>>>>>      656:    e9 00 00 00 00           jmpq   65b
>>>>> <drm_calc_timestamping_constants+0x12b>
>>>>>      65b:    41 8b 54 24 60           mov    0x60(%r12),%edx
>>>>>      660:    49 8b 7f 08              mov    0x8(%r15),%rdi
>>>>>      664:    44 89 45 c4              mov    %r8d,-0x3c(%rbp)
>>>>>      668:    45 31 ed                 xor    %r13d,%r13d
>>>>>      66b:    48 c7 c6 00 00 00 00     mov    $0x0,%rsi
>>>>>      672:    48 89 4d c8              mov    %rcx,-0x38(%rbp)
>>>>>      676:    e8 00 00 00 00           callq  67b
>>>>> <drm_calc_timestamping_constants+0x14b>
>>>>>      67b:    c7 45 d0 00 00 00 00     movl   $0x0,-0x30(%rbp)
>>>>>      682:    44 8b 45 c4              mov    -0x3c(%rbp),%r8d
>>>>>      686:    48 8b 4d c8              mov    -0x38(%rbp),%rcx
>>>>>      68a:    e9 39 ff ff ff           jmpq   5c8
>>>>> <drm_calc_timestamping_constants+0x98>
>>>>>      68f:    52                       push   %rdx
>>>>>      690:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>>>>      697:    31 d2                    xor    %edx,%edx
>>>>>      699:    31 f6                    xor    %esi,%esi
>>>>>      69b:    50                       push   %rax
>>>>>      69c:    31 ff                    xor    %edi,%edi
>>>>>      69e:    e8 00 00 00 00           callq  6a3
>>>>> <drm_calc_timestamping_constants+0x173>
>>>>>      6a3:    45 8b 44 24 60           mov    0x60(%r12),%r8d
>>>>>      6a8:    58                       pop    %rax
>>>>>      6a9:    5a                       pop    %rdx
>>>>>      6aa:    e9 7b ff ff ff           jmpq   62a
>>>>> <drm_calc_timestamping_constants+0xfa>
>>>>>      6af:    49 8b 7f 08              mov    0x8(%r15),%rdi
>>>>>      6b3:    4c 8b 67 50              mov    0x50(%rdi),%r12
>>>>>      6b7:    4d 85 e4                 test   %r12,%r12
>>>>>      6ba:    74 25                    je     6e1
>>>>> <drm_calc_timestamping_constants+0x1b1>
>>>>>      6bc:    e8 00 00 00 00           callq  6c1
>>>>> <drm_calc_timestamping_constants+0x191>
>>>>>      6c1:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>>>>      6c8:    4c 89 e2                 mov    %r12,%rdx
>>>>>      6cb:    48 c7 c7 00 00 00 00     mov    $0x0,%rdi
>>>>>      6d2:    48 89 c6                 mov    %rax,%rsi
>>>>>      6d5:    e8 00 00 00 00           callq  6da
>>>>> <drm_calc_timestamping_constants+0x1aa>
>>>>>      6da:    0f 0b                    ud2
>>>>>      6dc:    e9 67 ff ff ff           jmpq   648
>>>>> <drm_calc_timestamping_constants+0x118>
>>>>>      6e1:    4c 8b 27                 mov    (%rdi),%r12
>>>>>      6e4:    eb d6                    jmp    6bc
>>>>> <drm_calc_timestamping_constants+0x18c>
>>>>>      6e6:    66 2e 0f 1f 84 00 00     nopw   %cs:0x0(%rax,%rax,1)
>>>>>      6ed:    00 00 00
>>>>>      6f0:    90                       nop
>>>>>      6f1:    90                       nop
>>>>>      6f2:    90                       nop
>>>>>      6f3:    90                       nop
>>>>>      6f4:    90                       nop
>>>>>      6f5:    90                       nop
>>>>>      6f6:    90                       nop
>>>>>      6f7:    90                       nop
>>>>>      6f8:    90                       nop
>>>>>      6f9:    90                       nop
>>>>>      6fa:    90                       nop
>>>>>      6fb:    90                       nop
>>>>>      6fc:    90                       nop
>>>>>      6fd:    90                       nop
>>>>>      6fe:    90                       nop
>>>>>      6ff:    90                       nop
>>>>>
>>>>>
>>>>>> David
>>>>>>
>>>>>> -
>>>>>> Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
>>>>>> Registration No: 1397386 (Wales)
>>>>>>

--
Jani Nikula, Intel Open Source Graphics Center

2023-05-23 04:04:28

by Sui Jingfeng

[permalink] [raw]
Subject: Re: [PATCH] drm/drm_vblank.c: avoid unsigned int to signed int cast

Hi,

On 2023/5/22 23:01, Jani Nikula wrote:
> On Mon, 22 May 2023, Sui Jingfeng <[email protected]> wrote:
>> Hi,
>>
>> On 2023/5/22 20:13, Jani Nikula wrote:
>>> On Mon, 22 May 2023, Sui Jingfeng <[email protected]> wrote:
>>>> Hi,
>>>>
>>>> On 2023/5/22 19:29, Jani Nikula wrote:
>>>>> On Thu, 18 May 2023, Sui Jingfeng <[email protected]> wrote:
>>>>>> On 2023/5/17 18:59, David Laight wrote:
>>>>>>> From: [email protected]
>>>>>>>> Sent: 16 May 2023 18:30
>>>>>>>>
>>>>>>>> From: Sui Jingfeng <[email protected]>
>>>>>>>>
>>>>>>>> Both mode->crtc_htotal and mode->crtc_vtotal are u16 type,
>>>>>>>> mode->crtc_htotal * mode->crtc_vtotal will results a unsigned type.
>>>>>>> Nope, u16 gets promoted to 'signed int' and the result of the
>>>>>>> multiply is also signed.
>>>>>> I believe that signed or unsigned is dependent on the declaration.
>>>>>>
>>>>>> I am talk about the math, while you are talking about compiler.
>>>>>>
>>>>>> I admit that u16 gets promoted to 'signed int' is true, but this is
>>>>>> irrelevant,
>>>>>>
>>>>>> the point is how to understand the returned value.
>>>>>>
>>>>>>
>>>>>> How does the compiler generate the code is one thing, how do we
>>>>>> interpret the result is another
>>>>>>
>>>>>> How does the compiler generate the code is NOT determined by us, while
>>>>>> how do we interpret the result is determined by us.
>>>>>>
>>>>>>
>>>>>> I believe that using a u32 type to interpret the result(u16 * u16) is
>>>>>> always true, it is true in the perspective of *math*.
>>>>>>
>>>>>> Integer promotions is the details of C program language. If the result
>>>>>> of the multiply is signed, then there are risks that
>>>>>>
>>>>>> the result is negative, what's the benefit to present this risk to the
>>>>>> programmer?
>>>>>>
>>>>>> What's the benefit to tell me(and others) that u16 * u16 yield a signed
>>>>>> value? and can be negative?
>>>>>>
>>>>>> Using int type as the return type bring concerns to the programmer and
>>>>>> the user of the function,
>>>>>>
>>>>>> even though this is not impossible in practice.
>>>>> In general, do not use unsigned types in arithmethic to avoid negative
>>>>> values, because most people will be tripped over by integer promotion
>>>>> rules, and you'll get negative values anyway.
>>>>>
>>>>> I'll bet most people will be surprised to see what this prints:
>>>>>
>>>>> #include <stdio.h>
>>>>> #include <stdint.h>
>>>>>
>>>>> int main(void)
>>>>> {
>>>>> uint16_t x = 0xffff;
>>>>> uint16_t y = 0xffff;
>>>>> uint64_t z = x * y;
>>>>>
>>>>> printf("0x%016lx\n", z);
>>>>> printf("%ld\n", z);
>>>> Here, please replace the "%ld\n" with the "%lu\n", then you will see the
>>>> difference.
>>>>
>>>> you are casting the variable 'z' to signed value,  "%d" is for printing
>>>> signed value, and "%u" is for printing unsigned value.
>>>>
>>>>
>>>> Your simple code explained exactly why you are still in confusion,
>>> Am I?
>>>
>>> Take a look at the values, and explain the math.
>> I meant the value itself is represent with 2's compliment,
>>
>> when you print a value with '%ld', then you will get the signed version,
>>
>> when you print a value with '%lu', then you will get the unsigned version.
>>
>> The result of a u16*u16 couldn't be negative in math.
>>
>>
>> But when you using a '%ld' or '%d' to print a unsigned value, then is wrong.
>>
>> This is also the case which you shouldn't using a int type to store the result of u16*u16.
>>
>> because when I seen a int type, I will choose '%d' to print it,
>>
>> when I seen a unsigned int type, I will choose '%u' to print it.
>>
>> when using a int type as the return type, this could lead people to using '%d' to print
>>
>> such a value. Then, it generate the confusion as this little test program shows.
> Using 0x%016lx and %lu results in 0xfffffffffffe0001 and
> 18446744073709420545, respectively. They are equal. They are indeed not
> negative.
>
> However 0xffff * 0xffff = 0xfffe0001. Or 4294836225 in decimal.
>
> No matter what the math says, this is what actually happens in C.
>
> I don't know what more I could possibly tell you.

Sorry, I realized something after rethink about it.

Can we first assign the value to u32 first, then expend it to 64 bit then?

Extend it to 64 bit from 32 bit explicitly, this enforce zero extend
instead of sign extend.

>
> BR,
> Jani.
>
>
>>> BR,
>>> Jani.
>>>
>>>> that is u16 * u16  can yield a negative value if you use the int as the
>>>> return type. Because it overflowed.
>>>>
>>>>> printf("%d\n", x * y);
>>>>> }
>>>>>
>>>>> And it's not that different from what you have below. Your patch doesn't
>>>>> change anything, and doesn't make it any less confusing.
>>>>>
>>>>> BR,
>>>>> Jani.
>>>>>
>>>>>
>>>>>>>> Using a u32 is enough to store the result, but considering that the
>>>>>>>> result will be casted to u64 soon after. We use a u64 type directly.
>>>>>>>> So there no need to cast it to signed type and cast back then.
>>>>>>> ....
>>>>>>>> - int frame_size = mode->crtc_htotal * mode->crtc_vtotal;
>>>>>>>> + u64 frame_size = mode->crtc_htotal * mode->crtc_vtotal;
>>>>>>> ...
>>>>>>>> - framedur_ns = div_u64((u64) frame_size * 1000000, dotclock);
>>>>>>>> + framedur_ns = div_u64(frame_size * 1000000, dotclock);
>>>>>>> The (u64) cast is there to extend the value to 64bits, not
>>>>>>> because the original type is signed.
>>>>>> Sorry about my expression, I think my sentence did not mention anything
>>>>>> about 'because the original type is signed'.
>>>>>>
>>>>>> In the contrary, my patch eliminated the concerns to the reviewer. It
>>>>>> say that the results of the multiply can't be negative.
>>>>>>
>>>>>> My intent is to tell the compiler we want a unsigned return type, but
>>>>>> GCC emit 'imul' instruction for the multiply......
>>>>>>
>>>>>> I'm using u64 as the return type, because div_u64() function accept a
>>>>>> u64 type value as its first argument.
>>>>>>
>>>>>>> The compiler will detect that the old code is a 32x32 multiply
>>>>>>> where a 64bit result is needed, that may not be true for the
>>>>>>> changed code (it would need to track back as far as the u16s).
>>>>>> I don't believe my code could be wrong.
>>>>>>
>>>>>> when you use the word 'may', you are saying that it could be wrong after
>>>>>> apply my patch.
>>>>>>
>>>>>> Then you have to find at least one test example to prove you point, in
>>>>>> which case my codes generate wrong results.
>>>>>>
>>>>>> Again I don't believe you could find one.
>>>>>>
>>>>>>> It is not uncommon to force a 64bit result from a multiply
>>>>>>> by making the constant 64bit. As in:
>>>>>>> div_u64(frame_size * 1000000ULL, dotclock);
>>>>>> In fact, After apply this patch, the ASM code generated is same with before.
>>>>>>
>>>>>> This may because the GCC is smart enough to generate optimized code in
>>>>>> either case,
>>>>>>
>>>>>> I think It could be different with a different optimization-level.
>>>>>>
>>>>>> I have tested this patch on three different architecture,  I can not
>>>>>> find error still.
>>>>>>
>>>>>> Below is the assembly extract on x86-64: because GCC generate the same
>>>>>> code in either case,
>>>>>>
>>>>>> so I pasted only one copy here.
>>>>>>
>>>>>>
>>>>>> 0000000000000530 <drm_calc_timestamping_constants>:
>>>>>>      530:    f3 0f 1e fa              endbr64
>>>>>>      534:    e8 00 00 00 00           callq  539
>>>>>> <drm_calc_timestamping_constants+0x9>
>>>>>>      539:    55                       push   %rbp
>>>>>>      53a:    48 89 e5                 mov    %rsp,%rbp
>>>>>>      53d:    41 57                    push   %r15
>>>>>>      53f:    41 56                    push   %r14
>>>>>>      541:    41 55                    push   %r13
>>>>>>      543:    41 54                    push   %r12
>>>>>>      545:    53                       push   %rbx
>>>>>>      546:    48 83 ec 18              sub    $0x18,%rsp
>>>>>>      54a:    4c 8b 3f                 mov    (%rdi),%r15
>>>>>>      54d:    41 8b 87 6c 01 00 00     mov    0x16c(%r15),%eax
>>>>>>      554:    85 c0                    test   %eax,%eax
>>>>>>      556:    0f 84 ec 00 00 00        je     648
>>>>>> <drm_calc_timestamping_constants+0x118>
>>>>>>      55c:    44 8b 87 90 00 00 00     mov    0x90(%rdi),%r8d
>>>>>>      563:    49 89 fc                 mov    %rdi,%r12
>>>>>>      566:    44 39 c0                 cmp    %r8d,%eax
>>>>>>      569:    0f 86 40 01 00 00        jbe    6af
>>>>>> <drm_calc_timestamping_constants+0x17f>
>>>>>>      56f:    44 8b 76 1c              mov    0x1c(%rsi),%r14d
>>>>>>      573:    49 8b 8f 40 01 00 00     mov    0x140(%r15),%rcx
>>>>>>      57a:    48 89 f3                 mov    %rsi,%rbx
>>>>>>      57d:    45 85 f6                 test   %r14d,%r14d
>>>>>>      580:    0f 8e d5 00 00 00        jle    65b
>>>>>> <drm_calc_timestamping_constants+0x12b>
>>>>>>      586:    0f b7 43 2a              movzwl 0x2a(%rbx),%eax
>>>>>>      58a:    49 63 f6                 movslq %r14d,%rsi
>>>>>>      58d:    31 d2                    xor    %edx,%edx
>>>>>>      58f:    48 89 c7                 mov    %rax,%rdi
>>>>>>      592:    48 69 c0 40 42 0f 00     imul   $0xf4240,%rax,%rax
>>>>>>      599:    48 f7 f6                 div    %rsi
>>>>>>      59c:    31 d2                    xor    %edx,%edx
>>>>>>      59e:    48 89 45 d0              mov    %rax,-0x30(%rbp)
>>>>>>      5a2:    0f b7 43 38              movzwl 0x38(%rbx),%eax
>>>>>>      5a6:    0f af c7                 imul   %edi,%eax
>>>>>>      5a9:    48 98                    cltq
>>>>>>      5ab:    48 69 c0 40 42 0f 00     imul   $0xf4240,%rax,%rax
>>>>>>      5b2:    48 f7 f6                 div    %rsi
>>>>>>      5b5:    41 89 c5                 mov    %eax,%r13d
>>>>>>      5b8:    f6 43 18 10              testb  $0x10,0x18(%rbx)
>>>>>>      5bc:    74 0a                    je     5c8
>>>>>> <drm_calc_timestamping_constants+0x98>
>>>>>>      5be:    41 c1 ed 1f              shr    $0x1f,%r13d
>>>>>>      5c2:    41 01 c5                 add    %eax,%r13d
>>>>>>      5c5:    41 d1 fd                 sar    %r13d
>>>>>>      5c8:    4b 8d 04 c0              lea    (%r8,%r8,8),%rax
>>>>>>      5cc:    48 89 de                 mov    %rbx,%rsi
>>>>>>      5cf:    49 8d 3c 40              lea    (%r8,%rax,2),%rdi
>>>>>>      5d3:    8b 45 d0                 mov    -0x30(%rbp),%eax
>>>>>>      5d6:    48 c1 e7 04              shl    $0x4,%rdi
>>>>>>      5da:    48 01 cf                 add    %rcx,%rdi
>>>>>>      5dd:    89 47 78                 mov    %eax,0x78(%rdi)
>>>>>>      5e0:    48 83 ef 80              sub $0xffffffffffffff80,%rdi
>>>>>>      5e4:    44 89 6f f4              mov    %r13d,-0xc(%rdi)
>>>>>>      5e8:    e8 00 00 00 00           callq  5ed
>>>>>> <drm_calc_timestamping_constants+0xbd>
>>>>>>      5ed:    0f b7 53 2e              movzwl 0x2e(%rbx),%edx
>>>>>>      5f1:    0f b7 43 38              movzwl 0x38(%rbx),%eax
>>>>>>      5f5:    44 0f b7 4b 2a           movzwl 0x2a(%rbx),%r9d
>>>>>>      5fa:    45 8b 44 24 60           mov    0x60(%r12),%r8d
>>>>>>      5ff:    4d 85 ff                 test   %r15,%r15
>>>>>>      602:    0f 84 87 00 00 00        je     68f
>>>>>> <drm_calc_timestamping_constants+0x15f>
>>>>>>      608:    49 8b 77 08              mov    0x8(%r15),%rsi
>>>>>>      60c:    52                       push   %rdx
>>>>>>      60d:    31 ff                    xor    %edi,%edi
>>>>>>      60f:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>>>>>      616:    50                       push   %rax
>>>>>>      617:    31 d2                    xor    %edx,%edx
>>>>>>      619:    e8 00 00 00 00           callq  61e
>>>>>> <drm_calc_timestamping_constants+0xee>
>>>>>>      61e:    45 8b 44 24 60           mov    0x60(%r12),%r8d
>>>>>>      623:    4d 8b 7f 08              mov    0x8(%r15),%r15
>>>>>>      627:    5f                       pop    %rdi
>>>>>>      628:    41 59                    pop    %r9
>>>>>>      62a:    8b 45 d0                 mov    -0x30(%rbp),%eax
>>>>>>      62d:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>>>>>      634:    4c 89 fe                 mov    %r15,%rsi
>>>>>>      637:    45 89 f1                 mov    %r14d,%r9d
>>>>>>      63a:    31 d2                    xor    %edx,%edx
>>>>>>      63c:    31 ff                    xor    %edi,%edi
>>>>>>      63e:    50                       push   %rax
>>>>>>      63f:    41 55                    push   %r13
>>>>>>      641:    e8 00 00 00 00           callq  646
>>>>>> <drm_calc_timestamping_constants+0x116>
>>>>>>      646:    59                       pop    %rcx
>>>>>>      647:    5e                       pop    %rsi
>>>>>>      648:    48 8d 65 d8              lea    -0x28(%rbp),%rsp
>>>>>>      64c:    5b                       pop    %rbx
>>>>>>      64d:    41 5c                    pop    %r12
>>>>>>      64f:    41 5d                    pop    %r13
>>>>>>      651:    41 5e                    pop    %r14
>>>>>>      653:    41 5f                    pop    %r15
>>>>>>      655:    5d                       pop    %rbp
>>>>>>      656:    e9 00 00 00 00           jmpq   65b
>>>>>> <drm_calc_timestamping_constants+0x12b>
>>>>>>      65b:    41 8b 54 24 60           mov    0x60(%r12),%edx
>>>>>>      660:    49 8b 7f 08              mov    0x8(%r15),%rdi
>>>>>>      664:    44 89 45 c4              mov    %r8d,-0x3c(%rbp)
>>>>>>      668:    45 31 ed                 xor    %r13d,%r13d
>>>>>>      66b:    48 c7 c6 00 00 00 00     mov    $0x0,%rsi
>>>>>>      672:    48 89 4d c8              mov    %rcx,-0x38(%rbp)
>>>>>>      676:    e8 00 00 00 00           callq  67b
>>>>>> <drm_calc_timestamping_constants+0x14b>
>>>>>>      67b:    c7 45 d0 00 00 00 00     movl   $0x0,-0x30(%rbp)
>>>>>>      682:    44 8b 45 c4              mov    -0x3c(%rbp),%r8d
>>>>>>      686:    48 8b 4d c8              mov    -0x38(%rbp),%rcx
>>>>>>      68a:    e9 39 ff ff ff           jmpq   5c8
>>>>>> <drm_calc_timestamping_constants+0x98>
>>>>>>      68f:    52                       push   %rdx
>>>>>>      690:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>>>>>      697:    31 d2                    xor    %edx,%edx
>>>>>>      699:    31 f6                    xor    %esi,%esi
>>>>>>      69b:    50                       push   %rax
>>>>>>      69c:    31 ff                    xor    %edi,%edi
>>>>>>      69e:    e8 00 00 00 00           callq  6a3
>>>>>> <drm_calc_timestamping_constants+0x173>
>>>>>>      6a3:    45 8b 44 24 60           mov    0x60(%r12),%r8d
>>>>>>      6a8:    58                       pop    %rax
>>>>>>      6a9:    5a                       pop    %rdx
>>>>>>      6aa:    e9 7b ff ff ff           jmpq   62a
>>>>>> <drm_calc_timestamping_constants+0xfa>
>>>>>>      6af:    49 8b 7f 08              mov    0x8(%r15),%rdi
>>>>>>      6b3:    4c 8b 67 50              mov    0x50(%rdi),%r12
>>>>>>      6b7:    4d 85 e4                 test   %r12,%r12
>>>>>>      6ba:    74 25                    je     6e1
>>>>>> <drm_calc_timestamping_constants+0x1b1>
>>>>>>      6bc:    e8 00 00 00 00           callq  6c1
>>>>>> <drm_calc_timestamping_constants+0x191>
>>>>>>      6c1:    48 c7 c1 00 00 00 00     mov    $0x0,%rcx
>>>>>>      6c8:    4c 89 e2                 mov    %r12,%rdx
>>>>>>      6cb:    48 c7 c7 00 00 00 00     mov    $0x0,%rdi
>>>>>>      6d2:    48 89 c6                 mov    %rax,%rsi
>>>>>>      6d5:    e8 00 00 00 00           callq  6da
>>>>>> <drm_calc_timestamping_constants+0x1aa>
>>>>>>      6da:    0f 0b                    ud2
>>>>>>      6dc:    e9 67 ff ff ff           jmpq   648
>>>>>> <drm_calc_timestamping_constants+0x118>
>>>>>>      6e1:    4c 8b 27                 mov    (%rdi),%r12
>>>>>>      6e4:    eb d6                    jmp    6bc
>>>>>> <drm_calc_timestamping_constants+0x18c>
>>>>>>      6e6:    66 2e 0f 1f 84 00 00     nopw   %cs:0x0(%rax,%rax,1)
>>>>>>      6ed:    00 00 00
>>>>>>      6f0:    90                       nop
>>>>>>      6f1:    90                       nop
>>>>>>      6f2:    90                       nop
>>>>>>      6f3:    90                       nop
>>>>>>      6f4:    90                       nop
>>>>>>      6f5:    90                       nop
>>>>>>      6f6:    90                       nop
>>>>>>      6f7:    90                       nop
>>>>>>      6f8:    90                       nop
>>>>>>      6f9:    90                       nop
>>>>>>      6fa:    90                       nop
>>>>>>      6fb:    90                       nop
>>>>>>      6fc:    90                       nop
>>>>>>      6fd:    90                       nop
>>>>>>      6fe:    90                       nop
>>>>>>      6ff:    90                       nop
>>>>>>
>>>>>>
>>>>>>> David
>>>>>>>
>>>>>>> -
>>>>>>> Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
>>>>>>> Registration No: 1397386 (Wales)
>>>>>>>

2023-05-23 05:05:59

by Sui Jingfeng

[permalink] [raw]
Subject: Re: [PATCH] drm/drm_vblank.c: avoid unsigned int to signed int cast

Hi,

On 2023/5/22 19:29, Jani Nikula wrote:
> In general, do not use unsigned types in arithmethic to avoid negative
> values, because most people will be tripped over by integer promotion
> rules, and you'll get negative values anyway.


Here I'm sure about this,

but there are plenty unsigned types arithmetic in the kernel.

take kmalloc_array() function as an example in /tools/virto/linux/kernel.h


static inline void *kmalloc_array(unsigned n, size_t s, gfp_t gfp)
{
    return kmalloc(n * s, gfp);
}


NOTE that *size_t* is an unsigned integral data type.


2023-05-23 08:17:53

by Sui Jingfeng

[permalink] [raw]
Subject: Re: [PATCH] drm/drm_vblank.c: avoid unsigned int to signed int cast

Hi,

On 2023/5/23 12:26, Sui Jingfeng wrote:
> Hi,
>
> On 2023/5/22 19:29, Jani Nikula wrote:
>> In general, do not use unsigned types in arithmethic to avoid negative
>> values, because most people will be tripped over by integer promotion
>> rules, and you'll get negative values anyway.
>
>
> Here I'm sure about this,
>
Here, I'm NOT sure about this


> but there are plenty unsigned types arithmetic in the kernel.
>
> take kmalloc_array() function as an example in
> /tools/virto/linux/kernel.h
>
>
> static inline void *kmalloc_array(unsigned n, size_t s, gfp_t gfp)
> {
>     return kmalloc(n * s, gfp);
> }
>
>
> NOTE that *size_t* is an unsigned integral data type.

2023-05-23 08:57:50

by David Laight

[permalink] [raw]
Subject: RE: [PATCH] drm/drm_vblank.c: avoid unsigned int to signed int cast

From: [email protected] <[email protected]>
> Sent: 23 May 2023 05:27
>
> On 2023/5/22 19:29, Jani Nikula wrote:
> > In general, do not use unsigned types in arithmethic to avoid negative
> > values, because most people will be tripped over by integer promotion
> > rules, and you'll get negative values anyway.
>
>
> Here I'm sure about this,
>
> but there are plenty unsigned types arithmetic in the kernel.

The real problem is (attempted) arithmetic on types smaller than int.
Regardless of whether they are signed or unsigned.

David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)

2023-05-23 10:27:48

by Sui Jingfeng

[permalink] [raw]
Subject: Re: [PATCH] drm/drm_vblank.c: avoid unsigned int to signed int cast

Hi,

On 2023/5/23 16:50, David Laight wrote:
> From: [email protected] <[email protected]>
>> Sent: 23 May 2023 05:27
>>
>> On 2023/5/22 19:29, Jani Nikula wrote:
>>> In general, do not use unsigned types in arithmethic to avoid negative
>>> values, because most people will be tripped over by integer promotion
>>> rules, and you'll get negative values anyway.
>>
>> Here I'm not sure about this,
>>
>> but there are plenty unsigned types arithmetic in the kernel.
> The real problem is (attempted) arithmetic on types smaller than int.
> Regardless of whether they are signed or unsigned.

It is about sign extend.

Yes, you may be right. I might create a wrong patch.

But this will not happen in practice, because in general case:

mode->crtc_htotal < 0x8fff;

mode->crtc_vtotal < 0x8fff;

u16 gets promoted to 'signed int' not 'unsigned int'.

Sorry  :/

> David
>
> -
> Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
> Registration No: 1397386 (Wales)