2023-05-04 20:41:14

by Sasha Levin

Subject: [PATCH AUTOSEL 6.3 08/59] bpf, mips: Implement DADDI workarounds for JIT

From: Jiaxun Yang <[email protected]>

[ Upstream commit bbefef2f07080cd502a93cb1c529e1c8a6c4ac8e ]

For DADDI errata we just workaround by disable immediate operation
for BPF_ADD / BPF_SUB to avoid generation of DADDIU.

All other use cases in JIT won't cause overflow thus they are all safe.

Signed-off-by: Jiaxun Yang <[email protected]>
Signed-off-by: Daniel Borkmann <[email protected]>
Reviewed-by: Philippe Mathieu-Daudé <[email protected]>
Acked-by: Johan Almbladh <[email protected]>
Link: https://lore.kernel.org/bpf/[email protected]
Signed-off-by: Sasha Levin <[email protected]>
---
arch/mips/Kconfig | 1 -
arch/mips/net/bpf_jit_comp.c | 4 ++++
2 files changed, 4 insertions(+), 1 deletion(-)

diff --git a/arch/mips/Kconfig b/arch/mips/Kconfig
index e2f3ca73f40d6..edc7d8790f1e8 100644
--- a/arch/mips/Kconfig
+++ b/arch/mips/Kconfig
@@ -64,7 +64,6 @@ config MIPS
select HAVE_DMA_CONTIGUOUS
select HAVE_DYNAMIC_FTRACE
select HAVE_EBPF_JIT if !CPU_MICROMIPS && \
- !CPU_DADDI_WORKAROUNDS && \
!CPU_R4000_WORKAROUNDS && \
!CPU_R4400_WORKAROUNDS
select HAVE_EXIT_THREAD
diff --git a/arch/mips/net/bpf_jit_comp.c b/arch/mips/net/bpf_jit_comp.c
index b17130d510d49..a40d926b65139 100644
--- a/arch/mips/net/bpf_jit_comp.c
+++ b/arch/mips/net/bpf_jit_comp.c
@@ -218,9 +218,13 @@ bool valid_alu_i(u8 op, s32 imm)
/* All legal eBPF values are valid */
return true;
case BPF_ADD:
+ if (IS_ENABLED(CONFIG_CPU_DADDI_WORKAROUNDS))
+ return false;
/* imm must be 16 bits */
return imm >= -0x8000 && imm <= 0x7fff;
case BPF_SUB:
+ if (IS_ENABLED(CONFIG_CPU_DADDI_WORKAROUNDS))
+ return false;
/* -imm must be 16 bits */
return imm >= -0x7fff && imm <= 0x8000;
case BPF_AND:
--
2.39.2

2023-05-05 12:10:31

by Jiaxun Yang

Subject: Re: [PATCH AUTOSEL 6.3 08/59] bpf, mips: Implement DADDI workarounds for JIT

> 2023年5月4日 20:40，Sasha Levin <[email protected]> 写道：
>
> From: Jiaxun Yang <[email protected]>
>
> [ Upstream commit bbefef2f07080cd502a93cb1c529e1c8a6c4ac8e ]
>
> For DADDI errata we just workaround by disable immediate operation
> for BPF_ADD / BPF_SUB to avoid generation of DADDIU.
>
> All other use cases in JIT won't cause overflow thus they are all safe.
>
> Signed-off-by: Jiaxun Yang <[email protected]>
> Signed-off-by: Daniel Borkmann <[email protected]>
> Reviewed-by: Philippe Mathieu-Daudé <[email protected]>
> Acked-by: Johan Almbladh <[email protected]>
> Link: https://lore.kernel.org/bpf/[email protected]
> Signed-off-by: Sasha Levin <[email protected]>

Hi Sasha,

I think this patch should count as a functional improvement instead of regression fix.

Please drop it from stable queue.

Thanks
Jiaxun

> ---
> arch/mips/Kconfig | 1 -
> arch/mips/net/bpf_jit_comp.c | 4 ++++
> 2 files changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/arch/mips/Kconfig b/arch/mips/Kconfig
> index e2f3ca73f40d6..edc7d8790f1e8 100644
> --- a/arch/mips/Kconfig
> +++ b/arch/mips/Kconfig
> @@ -64,7 +64,6 @@ config MIPS
> select HAVE_DMA_CONTIGUOUS
> select HAVE_DYNAMIC_FTRACE
> select HAVE_EBPF_JIT if !CPU_MICROMIPS && \
> - !CPU_DADDI_WORKAROUNDS && \
> !CPU_R4000_WORKAROUNDS && \
> !CPU_R4400_WORKAROUNDS
> select HAVE_EXIT_THREAD
> diff --git a/arch/mips/net/bpf_jit_comp.c b/arch/mips/net/bpf_jit_comp.c
> index b17130d510d49..a40d926b65139 100644
> --- a/arch/mips/net/bpf_jit_comp.c
> +++ b/arch/mips/net/bpf_jit_comp.c
> @@ -218,9 +218,13 @@ bool valid_alu_i(u8 op, s32 imm)
> /* All legal eBPF values are valid */
> return true;
> case BPF_ADD:
> + if (IS_ENABLED(CONFIG_CPU_DADDI_WORKAROUNDS))
> + return false;
> /* imm must be 16 bits */
> return imm >= -0x8000 && imm <= 0x7fff;
> case BPF_SUB:
> + if (IS_ENABLED(CONFIG_CPU_DADDI_WORKAROUNDS))
> + return false;
> /* -imm must be 16 bits */
> return imm >= -0x7fff && imm <= 0x8000;
> case BPF_AND:
> --
> 2.39.2
>

2023-05-18 17:08:20

by Sasha Levin

Subject: Re: [PATCH AUTOSEL 6.3 08/59] bpf, mips: Implement DADDI workarounds for JIT

On Fri, May 05, 2023 at 01:04:14PM +0100, Jiaxun Yang wrote:
>
>
>> 2023年5月4日 20:40，Sasha Levin <[email protected]> 写道：
>>
>> From: Jiaxun Yang <[email protected]>
>>
>> [ Upstream commit bbefef2f07080cd502a93cb1c529e1c8a6c4ac8e ]
>>
>> For DADDI errata we just workaround by disable immediate operation
>> for BPF_ADD / BPF_SUB to avoid generation of DADDIU.
>>
>> All other use cases in JIT won't cause overflow thus they are all safe.
>>
>> Signed-off-by: Jiaxun Yang <[email protected]>
>> Signed-off-by: Daniel Borkmann <[email protected]>
>> Reviewed-by: Philippe Mathieu-Daudé <[email protected]>
>> Acked-by: Johan Almbladh <[email protected]>
>> Link: https://lore.kernel.org/bpf/[email protected]
>> Signed-off-by: Sasha Levin <[email protected]>
>
>Hi Sasha,
>
>I think this patch should count as a functional improvement instead of regression fix.
>
>Please drop it from stable queue.

Dropped, thanks!

--
Thanks,
Sasha