x86/crash: fix crash_setup_memmap_entries() KASAN vmalloc-out-of-bounds gripe
[ 15.428011] BUG: KASAN: vmalloc-out-of-bounds in crash_setup_memmap_entries+0x17e/0x3a0
[ 15.428018] Write of size 8 at addr ffffc90000426008 by task kexec/1187
(gdb) list *crash_setup_memmap_entries+0x17e
0xffffffff8107cafe is in crash_setup_memmap_entries (arch/x86/kernel/crash.c:322).
317 unsigned long long mend)
318 {
319 unsigned long start, end;
320
321 cmem->ranges[0].start = mstart;
322 cmem->ranges[0].end = mend;
323 cmem->nr_ranges = 1;
324
325 /* Exclude elf header region */
326 start = image->arch.elf_load_addr;
(gdb)
We're excluding two ranges, allocate the scratch space we need to do that.
Signed-off-by: Mike Galbraith <[email protected]>
---
arch/x86/kernel/crash.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
--- a/arch/x86/kernel/crash.c
+++ b/arch/x86/kernel/crash.c
@@ -337,7 +337,7 @@ int crash_setup_memmap_entries(struct ki
struct crash_memmap_data cmd;
struct crash_mem *cmem;
- cmem = vzalloc(sizeof(struct crash_mem));
+ cmem = vzalloc(sizeof(struct crash_mem)+(2*sizeof(struct crash_mem_range)));
if (!cmem)
return -ENOMEM;
Hi Mike,
Thanks for the patch! I suggest always cc kexec list for kexec/kdump
patches.
On 04/15/21 at 07:56pm, Mike Galbraith wrote:
> x86/crash: fix crash_setup_memmap_entries() KASAN vmalloc-out-of-bounds gripe
>
> [ 15.428011] BUG: KASAN: vmalloc-out-of-bounds in crash_setup_memmap_entries+0x17e/0x3a0
> [ 15.428018] Write of size 8 at addr ffffc90000426008 by task kexec/1187
>
> (gdb) list *crash_setup_memmap_entries+0x17e
> 0xffffffff8107cafe is in crash_setup_memmap_entries (arch/x86/kernel/crash.c:322).
> 317 unsigned long long mend)
> 318 {
> 319 unsigned long start, end;
> 320
> 321 cmem->ranges[0].start = mstart;
> 322 cmem->ranges[0].end = mend;
> 323 cmem->nr_ranges = 1;
> 324
> 325 /* Exclude elf header region */
> 326 start = image->arch.elf_load_addr;
> (gdb)
>
> We're excluding two ranges, allocate the scratch space we need to do that.
I think 1 range should be fine, have you tested 1?
The code is just excluding the elf header space which will be loaded
first before anything else so I assume it will be just at the start of
the crashkernel resource region. Thus [a b] after exclude the start
part will be [c b]. But I have not read the code for long time, maybe I
need to double check.
But anyway 2 would be good since the code is obscure we can easily miss
it in the future. See how other people think.
>
> Signed-off-by: Mike Galbraith <[email protected]>
> ---
> arch/x86/kernel/crash.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> --- a/arch/x86/kernel/crash.c
> +++ b/arch/x86/kernel/crash.c
> @@ -337,7 +337,7 @@ int crash_setup_memmap_entries(struct ki
> struct crash_memmap_data cmd;
> struct crash_mem *cmem;
>
> - cmem = vzalloc(sizeof(struct crash_mem));
> + cmem = vzalloc(sizeof(struct crash_mem)+(2*sizeof(struct crash_mem_range)));
Thanks for the patch, can you try below?
vzalloc(struct_size(cmem, ranges, 2));
> if (!cmem)
> return -ENOMEM;
>
>
Thanks
Dave
On Fri, 2021-04-16 at 19:07 +0800, Dave Young wrote:
>
> > We're excluding two ranges, allocate the scratch space we need to do that.
>
> I think 1 range should be fine, have you tested 1?
Have now, and vzalloc(struct_size(cmem, ranges, 1)) worked just fine.
-Mike
On 04/16/21 at 01:28pm, Mike Galbraith wrote:
> On Fri, 2021-04-16 at 19:07 +0800, Dave Young wrote:
> >
> > > We're excluding two ranges, allocate the scratch space we need to do that.
> >
> > I think 1 range should be fine, have you tested 1?
>
> Have now, and vzalloc(struct_size(cmem, ranges, 1)) worked just fine.
Ok, thanks for your quick response. Care to resend and cc x86 list and
Andrew?
Andrew usually takes core kexec/kdump fixes, x86 usually go through x86
maintainer.
Thanks
Dave
[ 15.428011] BUG: KASAN: vmalloc-out-of-bounds in crash_setup_memmap_entries+0x17e/0x3a0
[ 15.428018] Write of size 8 at addr ffffc90000426008 by task kexec/1187
(gdb) list *crash_setup_memmap_entries+0x17e
0xffffffff8107cafe is in crash_setup_memmap_entries (arch/x86/kernel/crash.c:322).
317 unsigned long long mend)
318 {
319 unsigned long start, end;
320
321 cmem->ranges[0].start = mstart;
322 cmem->ranges[0].end = mend;
323 cmem->nr_ranges = 1;
324
325 /* Exclude elf header region */
326 start = image->arch.elf_load_addr;
(gdb)
Append missing struct crash_mem_range to cmem.
Signed-off-by: Mike Galbraith <[email protected]>
---
arch/x86/kernel/crash.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
--- a/arch/x86/kernel/crash.c
+++ b/arch/x86/kernel/crash.c
@@ -337,7 +337,7 @@ int crash_setup_memmap_entries(struct ki
struct crash_memmap_data cmd;
struct crash_mem *cmem;
- cmem = vzalloc(sizeof(struct crash_mem));
+ cmem = vzalloc(struct_size(cmem, ranges, 1));
if (!cmem)
return -ENOMEM;
On Fri, Apr 16, 2021 at 02:02:07PM +0200, Mike Galbraith wrote:
> [ 15.428011] BUG: KASAN: vmalloc-out-of-bounds in crash_setup_memmap_entries+0x17e/0x3a0
> [ 15.428018] Write of size 8 at addr ffffc90000426008 by task kexec/1187
>
> (gdb) list *crash_setup_memmap_entries+0x17e
> 0xffffffff8107cafe is in crash_setup_memmap_entries (arch/x86/kernel/crash.c:322).
> 317 unsigned long long mend)
> 318 {
> 319 unsigned long start, end;
> 320
> 321 cmem->ranges[0].start = mstart;
> 322 cmem->ranges[0].end = mend;
> 323 cmem->nr_ranges = 1;
> 324
> 325 /* Exclude elf header region */
> 326 start = image->arch.elf_load_addr;
> (gdb)
>
> Append missing struct crash_mem_range to cmem.
This is winning this year's contest for most laconic patch commit
message! :-)
Please be more verbose and structure your commit message like this:
Problem is A.
It happens because of B.
Fix it by doing C.
(Potentially do D).
For more detailed info, see
Documentation/process/submitting-patches.rst, Section "2) Describe your
changes".
Thx.
--
Regards/Gruss,
Boris.
https://people.kernel.org/tglx/notes-about-netiquette
On Fri, 2021-04-16 at 16:44 +0200, Borislav Petkov wrote:
> On Fri, Apr 16, 2021 at 03:16:07PM +0200, Mike Galbraith wrote:
> > On Fri, 2021-04-16 at 14:16 +0200, Borislav Petkov wrote:
> > >
> > > Please be more verbose and structure your commit message like this:
> >
> > Hrmph, I thought it was too verbose for dinky one-liner if anything.
>
> Please look at how other commit messages in tip have free text - not
> only tools output.
>
> Also, this looks like a fix for some previous commit. Please dig out
> which commit introduced the issue and put its commit ID in a Fixes: tag
> above your S-o-B.
>
> If you don't have time or desire to do that, you can say so and I'll do
> it myself when I get a chance.
Ok, bin it for the nonce.
-Mike
On Fri, Apr 16 2021 at 17:13, Mike Galbraith wrote:
> On Fri, 2021-04-16 at 16:44 +0200, Borislav Petkov wrote:
>> On Fri, Apr 16, 2021 at 03:16:07PM +0200, Mike Galbraith wrote:
>> > On Fri, 2021-04-16 at 14:16 +0200, Borislav Petkov wrote:
>> > >
>> > > Please be more verbose and structure your commit message like this:
>> >
>> > Hrmph, I thought it was too verbose for dinky one-liner if anything.
>>
>> Please look at how other commit messages in tip have free text - not
>> only tools output.
>>
>> Also, this looks like a fix for some previous commit. Please dig out
>> which commit introduced the issue and put its commit ID in a Fixes: tag
>> above your S-o-B.
>>
>> If you don't have time or desire to do that, you can say so and I'll do
>> it myself when I get a chance.
>
> Ok, bin it for the nonce.
Can all of you involved stop this sandpit fight and do something useful
to fix that obvious bug already?
OMG
On Fri, 2021-04-16 at 23:44 +0200, Thomas Gleixner wrote:
>
> Can all of you involved stop this sandpit fight and do something useful
> to fix that obvious bug already?
?? We're not fighting afaik. Boris hated my changelog enough to offer
to write a better one, and I'm fine with that. It's a seven year old
*latent* buglet of microscopic proportions, hardly a pressing issue.
-Mike
Here's an attempt to explain what this fixes:
---
From: Mike Galbraith <[email protected]>
Date: Fri, 16 Apr 2021 14:02:07 +0200
Subject: [PATCH] x86/crash: Fix crash_setup_memmap_entries() out-of-bounds
access
Commit in Fixes: added support for kexec-ing a kernel on panic using a
new system call. As part of it, it does prepare a memory map for the new
kernel.
However, while doing so, it wrongly accesses memory it has not
allocated: it accesses the first element of the cmem->ranges[] array in
memmap_exclude_ranges() but it has not allocated the memory for it in
crash_setup_memmap_entries(). As KASAN reports:
BUG: KASAN: vmalloc-out-of-bounds in crash_setup_memmap_entries+0x17e/0x3a0
Write of size 8 at addr ffffc90000426008 by task kexec/1187
(gdb) list *crash_setup_memmap_entries+0x17e
0xffffffff8107cafe is in crash_setup_memmap_entries (arch/x86/kernel/crash.c:322).
317 unsigned long long mend)
318 {
319 unsigned long start, end;
320
321 cmem->ranges[0].start = mstart;
322 cmem->ranges[0].end = mend;
323 cmem->nr_ranges = 1;
324
325 /* Exclude elf header region */
326 start = image->arch.elf_load_addr;
(gdb)
Make sure the ranges array becomes a single element allocated.
[ bp: Write a proper commit message. ]
Fixes: dd5f726076cc ("kexec: support for kexec on panic using new system call")
Signed-off-by: Mike Galbraith <[email protected]>
Signed-off-by: Borislav Petkov <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
---
arch/x86/kernel/crash.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c
index a8f3af257e26..b1deacbeb266 100644
--- a/arch/x86/kernel/crash.c
+++ b/arch/x86/kernel/crash.c
@@ -337,7 +337,7 @@ int crash_setup_memmap_entries(struct kimage *image, struct boot_params *params)
struct crash_memmap_data cmd;
struct crash_mem *cmem;
- cmem = vzalloc(sizeof(struct crash_mem));
+ cmem = vzalloc(struct_size(cmem, ranges, 1));
if (!cmem)
return -ENOMEM;
--
2.29.2
--
Regards/Gruss,
Boris.
https://people.kernel.org/tglx/notes-about-netiquette
On 04/19/21 at 10:52am, Borislav Petkov wrote:
> Here's an attempt to explain what this fixes:
>
> ---
> From: Mike Galbraith <[email protected]>
> Date: Fri, 16 Apr 2021 14:02:07 +0200
> Subject: [PATCH] x86/crash: Fix crash_setup_memmap_entries() out-of-bounds
> access
>
> Commit in Fixes: added support for kexec-ing a kernel on panic using a
> new system call. As part of it, it does prepare a memory map for the new
> kernel.
>
> However, while doing so, it wrongly accesses memory it has not
> allocated: it accesses the first element of the cmem->ranges[] array in
> memmap_exclude_ranges() but it has not allocated the memory for it in
> crash_setup_memmap_entries(). As KASAN reports:
>
> BUG: KASAN: vmalloc-out-of-bounds in crash_setup_memmap_entries+0x17e/0x3a0
> Write of size 8 at addr ffffc90000426008 by task kexec/1187
>
> (gdb) list *crash_setup_memmap_entries+0x17e
> 0xffffffff8107cafe is in crash_setup_memmap_entries (arch/x86/kernel/crash.c:322).
> 317 unsigned long long mend)
> 318 {
> 319 unsigned long start, end;
> 320
> 321 cmem->ranges[0].start = mstart;
> 322 cmem->ranges[0].end = mend;
> 323 cmem->nr_ranges = 1;
> 324
> 325 /* Exclude elf header region */
> 326 start = image->arch.elf_load_addr;
> (gdb)
>
> Make sure the ranges array becomes a single element allocated.
>
> [ bp: Write a proper commit message. ]
Reviewed-by: Dave Young <[email protected]>
>
> Fixes: dd5f726076cc ("kexec: support for kexec on panic using new system call")
> Signed-off-by: Mike Galbraith <[email protected]>
> Signed-off-by: Borislav Petkov <[email protected]>
> Link: https://lkml.kernel.org/r/[email protected]
> ---
> arch/x86/kernel/crash.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c
> index a8f3af257e26..b1deacbeb266 100644
> --- a/arch/x86/kernel/crash.c
> +++ b/arch/x86/kernel/crash.c
> @@ -337,7 +337,7 @@ int crash_setup_memmap_entries(struct kimage *image, struct boot_params *params)
> struct crash_memmap_data cmd;
> struct crash_mem *cmem;
>
> - cmem = vzalloc(sizeof(struct crash_mem));
> + cmem = vzalloc(struct_size(cmem, ranges, 1));
> if (!cmem)
> return -ENOMEM;
>
> --
> 2.29.2
>
> --
> Regards/Gruss,
> Boris.
>
> https://people.kernel.org/tglx/notes-about-netiquette
>
Thanks
Dave
The following commit has been merged into the x86/urgent branch of tip:
Commit-ID: 5849cdf8c120e3979c57d34be55b92d90a77a47e
Gitweb: https://git.kernel.org/tip/5849cdf8c120e3979c57d34be55b92d90a77a47e
Author: Mike Galbraith <[email protected]>
AuthorDate: Fri, 16 Apr 2021 14:02:07 +02:00
Committer: Borislav Petkov <[email protected]>
CommitterDate: Tue, 20 Apr 2021 17:32:46 +02:00
x86/crash: Fix crash_setup_memmap_entries() out-of-bounds access
Commit in Fixes: added support for kexec-ing a kernel on panic using a
new system call. As part of it, it does prepare a memory map for the new
kernel.
However, while doing so, it wrongly accesses memory it has not
allocated: it accesses the first element of the cmem->ranges[] array in
memmap_exclude_ranges() but it has not allocated the memory for it in
crash_setup_memmap_entries(). As KASAN reports:
BUG: KASAN: vmalloc-out-of-bounds in crash_setup_memmap_entries+0x17e/0x3a0
Write of size 8 at addr ffffc90000426008 by task kexec/1187
(gdb) list *crash_setup_memmap_entries+0x17e
0xffffffff8107cafe is in crash_setup_memmap_entries (arch/x86/kernel/crash.c:322).
317 unsigned long long mend)
318 {
319 unsigned long start, end;
320
321 cmem->ranges[0].start = mstart;
322 cmem->ranges[0].end = mend;
323 cmem->nr_ranges = 1;
324
325 /* Exclude elf header region */
326 start = image->arch.elf_load_addr;
(gdb)
Make sure the ranges array becomes a single element allocated.
[ bp: Write a proper commit message. ]
Fixes: dd5f726076cc ("kexec: support for kexec on panic using new system call")
Signed-off-by: Mike Galbraith <[email protected]>
Signed-off-by: Borislav Petkov <[email protected]>
Reviewed-by: Dave Young <[email protected]>
Cc: <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
---
arch/x86/kernel/crash.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c
index a8f3af2..b1deacb 100644
--- a/arch/x86/kernel/crash.c
+++ b/arch/x86/kernel/crash.c
@@ -337,7 +337,7 @@ int crash_setup_memmap_entries(struct kimage *image, struct boot_params *params)
struct crash_memmap_data cmd;
struct crash_mem *cmem;
- cmem = vzalloc(sizeof(struct crash_mem));
+ cmem = vzalloc(struct_size(cmem, ranges, 1));
if (!cmem)
return -ENOMEM;