linuxppc-dev.lists.ozlabs.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] powerpc/ptrace: Simplify vr_get/set() to avoid GCC warning
@ 2019-02-15  6:14 Michael Ellerman
  2019-02-15  8:09 ` Mathieu Malaterre
                   ` (2 more replies)
  0 siblings, 3 replies; 6+ messages in thread
From: Michael Ellerman @ 2019-02-15  6:14 UTC (permalink / raw)
  To: linuxppc-dev; +Cc: malat, mroos

GCC 8 warns about the logic in vr_get/set(), which with -Werror breaks
the build:

  In function ‘user_regset_copyin’,
      inlined from ‘vr_set’ at arch/powerpc/kernel/ptrace.c:628:9:
  include/linux/regset.h:295:4: error: ‘memcpy’ offset [-527, -529] is
  out of the bounds [0, 16] of object ‘vrsave’ with type ‘union
  <anonymous>’ [-Werror=array-bounds]
  arch/powerpc/kernel/ptrace.c: In function ‘vr_set’:
  arch/powerpc/kernel/ptrace.c:623:5: note: ‘vrsave’ declared here
     } vrsave;

This has been identified as a regression in GCC, see GCC bug 88273.

However we can avoid the warning and also simplify the logic and make
it more robust.

Currently we pass -1 as end_pos to user_regset_copyout(). This says
"copy up to the end of the regset".

The definition of the regset is:
	[REGSET_VMX] = {
		.core_note_type = NT_PPC_VMX, .n = 34,
		.size = sizeof(vector128), .align = sizeof(vector128),
		.active = vr_active, .get = vr_get, .set = vr_set
	},

The end is calculated as (n * size), ie. 34 * sizeof(vector128).

In vr_get/set() we pass start_pos as 33 * sizeof(vector128), meaning
we can copy up to sizeof(vector128) into/out-of vrsave.

The on-stack vrsave is defined as:
  union {
	  elf_vrreg_t reg;
	  u32 word;
  } vrsave;

And elf_vrreg_t is:
  typedef __vector128 elf_vrreg_t;

So there is no bug, but we rely on all those sizes lining up,
otherwise we would have a kernel stack exposure/overwrite on our
hands.

Rather than relying on that we can pass an explict end_pos based on
the sizeof(vrsave). The result should be exactly the same but it's
more obviously not over-reading/writing the stack and it avoids the
compiler warning.

Reported-by: Meelis Roos <mroos@linux.ee>
Reported-by: Mathieu Malaterre <malat@debian.org>
Cc: stable@vger.kernel.org
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
---
 arch/powerpc/kernel/ptrace.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/arch/powerpc/kernel/ptrace.c b/arch/powerpc/kernel/ptrace.c
index 7535f89e08cd..d9ac7d94656e 100644
--- a/arch/powerpc/kernel/ptrace.c
+++ b/arch/powerpc/kernel/ptrace.c
@@ -567,6 +567,7 @@ static int vr_get(struct task_struct *target, const struct user_regset *regset,
 		/*
 		 * Copy out only the low-order word of vrsave.
 		 */
+		int start, end;
 		union {
 			elf_vrreg_t reg;
 			u32 word;
@@ -575,8 +576,10 @@ static int vr_get(struct task_struct *target, const struct user_regset *regset,
 
 		vrsave.word = target->thread.vrsave;
 
+		start = 33 * sizeof(vector128);
+		end = start + sizeof(vrsave);
 		ret = user_regset_copyout(&pos, &count, &kbuf, &ubuf, &vrsave,
-					  33 * sizeof(vector128), -1);
+					  start, end);
 	}
 
 	return ret;
@@ -614,6 +617,7 @@ static int vr_set(struct task_struct *target, const struct user_regset *regset,
 		/*
 		 * We use only the first word of vrsave.
 		 */
+		int start, end;
 		union {
 			elf_vrreg_t reg;
 			u32 word;
@@ -622,8 +626,10 @@ static int vr_set(struct task_struct *target, const struct user_regset *regset,
 
 		vrsave.word = target->thread.vrsave;
 
+		start = 33 * sizeof(vector128);
+		end = start + sizeof(vrsave);
 		ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, &vrsave,
-					 33 * sizeof(vector128), -1);
+					 start, end);
 		if (!ret)
 			target->thread.vrsave = vrsave.word;
 	}
-- 
2.20.1


^ permalink raw reply related	[flat|nested] 6+ messages in thread

* Re: [PATCH] powerpc/ptrace: Simplify vr_get/set() to avoid GCC warning
  2019-02-15  6:14 [PATCH] powerpc/ptrace: Simplify vr_get/set() to avoid GCC warning Michael Ellerman
@ 2019-02-15  8:09 ` Mathieu Malaterre
  2019-02-15  9:20   ` Michael Ellerman
  2019-02-16 19:15 ` Meelis Roos
  2019-02-22  9:47 ` Michael Ellerman
  2 siblings, 1 reply; 6+ messages in thread
From: Mathieu Malaterre @ 2019-02-15  8:09 UTC (permalink / raw)
  To: Michael Ellerman; +Cc: linuxppc-dev, Meelis Roos

On Fri, Feb 15, 2019 at 7:14 AM Michael Ellerman <mpe@ellerman.id.au> wrote:
>
> GCC 8 warns about the logic in vr_get/set(), which with -Werror breaks
> the build:
>
>   In function ‘user_regset_copyin’,
>       inlined from ‘vr_set’ at arch/powerpc/kernel/ptrace.c:628:9:
>   include/linux/regset.h:295:4: error: ‘memcpy’ offset [-527, -529] is
>   out of the bounds [0, 16] of object ‘vrsave’ with type ‘union
>   <anonymous>’ [-Werror=array-bounds]
>   arch/powerpc/kernel/ptrace.c: In function ‘vr_set’:
>   arch/powerpc/kernel/ptrace.c:623:5: note: ‘vrsave’ declared here
>      } vrsave;
>
> This has been identified as a regression in GCC, see GCC bug 88273.

Good point, this does not seems this will be backported.

> However we can avoid the warning and also simplify the logic and make
> it more robust.
>
> Currently we pass -1 as end_pos to user_regset_copyout(). This says
> "copy up to the end of the regset".
>
> The definition of the regset is:
>         [REGSET_VMX] = {
>                 .core_note_type = NT_PPC_VMX, .n = 34,
>                 .size = sizeof(vector128), .align = sizeof(vector128),
>                 .active = vr_active, .get = vr_get, .set = vr_set
>         },
>
> The end is calculated as (n * size), ie. 34 * sizeof(vector128).
>
> In vr_get/set() we pass start_pos as 33 * sizeof(vector128), meaning
> we can copy up to sizeof(vector128) into/out-of vrsave.
>
> The on-stack vrsave is defined as:
>   union {
>           elf_vrreg_t reg;
>           u32 word;
>   } vrsave;
>
> And elf_vrreg_t is:
>   typedef __vector128 elf_vrreg_t;
>
> So there is no bug, but we rely on all those sizes lining up,
> otherwise we would have a kernel stack exposure/overwrite on our
> hands.
>
> Rather than relying on that we can pass an explict end_pos based on
> the sizeof(vrsave). The result should be exactly the same but it's
> more obviously not over-reading/writing the stack and it avoids the
> compiler warning.
>

maybe:

Link: https://lkml.org/lkml/2018/8/16/117

In any case the warning is now gone:

Tested-by: Mathieu Malaterre <malat@debian.org>

> Reported-by: Meelis Roos <mroos@linux.ee>
> Reported-by: Mathieu Malaterre <malat@debian.org>
> Cc: stable@vger.kernel.org
> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
> ---
>  arch/powerpc/kernel/ptrace.c | 10 ++++++++--
>  1 file changed, 8 insertions(+), 2 deletions(-)
>
> diff --git a/arch/powerpc/kernel/ptrace.c b/arch/powerpc/kernel/ptrace.c
> index 7535f89e08cd..d9ac7d94656e 100644
> --- a/arch/powerpc/kernel/ptrace.c
> +++ b/arch/powerpc/kernel/ptrace.c
> @@ -567,6 +567,7 @@ static int vr_get(struct task_struct *target, const struct user_regset *regset,
>                 /*
>                  * Copy out only the low-order word of vrsave.
>                  */
> +               int start, end;
>                 union {
>                         elf_vrreg_t reg;
>                         u32 word;
> @@ -575,8 +576,10 @@ static int vr_get(struct task_struct *target, const struct user_regset *regset,
>
>                 vrsave.word = target->thread.vrsave;
>
> +               start = 33 * sizeof(vector128);
> +               end = start + sizeof(vrsave);
>                 ret = user_regset_copyout(&pos, &count, &kbuf, &ubuf, &vrsave,
> -                                         33 * sizeof(vector128), -1);
> +                                         start, end);
>         }
>
>         return ret;
> @@ -614,6 +617,7 @@ static int vr_set(struct task_struct *target, const struct user_regset *regset,
>                 /*
>                  * We use only the first word of vrsave.
>                  */
> +               int start, end;
>                 union {
>                         elf_vrreg_t reg;
>                         u32 word;
> @@ -622,8 +626,10 @@ static int vr_set(struct task_struct *target, const struct user_regset *regset,
>
>                 vrsave.word = target->thread.vrsave;
>
> +               start = 33 * sizeof(vector128);
> +               end = start + sizeof(vrsave);
>                 ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, &vrsave,
> -                                        33 * sizeof(vector128), -1);
> +                                        start, end);
>                 if (!ret)
>                         target->thread.vrsave = vrsave.word;
>         }
> --
> 2.20.1
>

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH] powerpc/ptrace: Simplify vr_get/set() to avoid GCC warning
  2019-02-15  8:09 ` Mathieu Malaterre
@ 2019-02-15  9:20   ` Michael Ellerman
  0 siblings, 0 replies; 6+ messages in thread
From: Michael Ellerman @ 2019-02-15  9:20 UTC (permalink / raw)
  To: Mathieu Malaterre; +Cc: linuxppc-dev, Meelis Roos

Mathieu Malaterre <malat@debian.org> writes:
> On Fri, Feb 15, 2019 at 7:14 AM Michael Ellerman <mpe@ellerman.id.au> wrote:
>>
>> GCC 8 warns about the logic in vr_get/set(), which with -Werror breaks
>> the build:
>>
>>   In function ‘user_regset_copyin’,
>>       inlined from ‘vr_set’ at arch/powerpc/kernel/ptrace.c:628:9:
>>   include/linux/regset.h:295:4: error: ‘memcpy’ offset [-527, -529] is
>>   out of the bounds [0, 16] of object ‘vrsave’ with type ‘union
>>   <anonymous>’ [-Werror=array-bounds]
>>   arch/powerpc/kernel/ptrace.c: In function ‘vr_set’:
>>   arch/powerpc/kernel/ptrace.c:623:5: note: ‘vrsave’ declared here
>>      } vrsave;
>>
>> This has been identified as a regression in GCC, see GCC bug 88273.
>
> Good point, this does not seems this will be backported.
>
>> However we can avoid the warning and also simplify the logic and make
>> it more robust.
>>
>> Currently we pass -1 as end_pos to user_regset_copyout(). This says
>> "copy up to the end of the regset".
>>
>> The definition of the regset is:
>>         [REGSET_VMX] = {
>>                 .core_note_type = NT_PPC_VMX, .n = 34,
>>                 .size = sizeof(vector128), .align = sizeof(vector128),
>>                 .active = vr_active, .get = vr_get, .set = vr_set
>>         },
>>
>> The end is calculated as (n * size), ie. 34 * sizeof(vector128).
>>
>> In vr_get/set() we pass start_pos as 33 * sizeof(vector128), meaning
>> we can copy up to sizeof(vector128) into/out-of vrsave.
>>
>> The on-stack vrsave is defined as:
>>   union {
>>           elf_vrreg_t reg;
>>           u32 word;
>>   } vrsave;
>>
>> And elf_vrreg_t is:
>>   typedef __vector128 elf_vrreg_t;
>>
>> So there is no bug, but we rely on all those sizes lining up,
>> otherwise we would have a kernel stack exposure/overwrite on our
>> hands.
>>
>> Rather than relying on that we can pass an explict end_pos based on
>> the sizeof(vrsave). The result should be exactly the same but it's
>> more obviously not over-reading/writing the stack and it avoids the
>> compiler warning.
>>
>
> maybe:
>
> Link: https://lkml.org/lkml/2018/8/16/117

Hmm, I prefer not to include links because they're unlikely to last
forever like the git history.

If we do include them the preferred form is a link to lkml.kernel.org
using the message id. That way the message is recorded and also the link
is "guaranteed" to work in future, eg:

http://lkml.kernel.org/r/alpine.LRH.2.21.1808161041350.16413@math.ut.ee

In this case I don't think the link adds much over what I have in the
change log, in particular I did credit Meelis as the reporter.

> In any case the warning is now gone:
>
> Tested-by: Mathieu Malaterre <malat@debian.org>

Thanks.

cheers

>> Reported-by: Meelis Roos <mroos@linux.ee>
>> Reported-by: Mathieu Malaterre <malat@debian.org>
>> Cc: stable@vger.kernel.org
>> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH] powerpc/ptrace: Simplify vr_get/set() to avoid GCC warning
  2019-02-15  6:14 [PATCH] powerpc/ptrace: Simplify vr_get/set() to avoid GCC warning Michael Ellerman
  2019-02-15  8:09 ` Mathieu Malaterre
@ 2019-02-16 19:15 ` Meelis Roos
  2019-02-19  1:02   ` Michael Ellerman
  2019-02-22  9:47 ` Michael Ellerman
  2 siblings, 1 reply; 6+ messages in thread
From: Meelis Roos @ 2019-02-16 19:15 UTC (permalink / raw)
  To: Michael Ellerman, linuxppc-dev; +Cc: malat

> Rather than relying on that we can pass an explict end_pos based on
> the sizeof(vrsave). The result should be exactly the same but it's
> more obviously not over-reading/writing the stack and it avoids the
> compiler warning.

It works on my PowerMac G4 with Debian-ports unstable with gcc 8.

-- 
Meelis Roos

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH] powerpc/ptrace: Simplify vr_get/set() to avoid GCC warning
  2019-02-16 19:15 ` Meelis Roos
@ 2019-02-19  1:02   ` Michael Ellerman
  0 siblings, 0 replies; 6+ messages in thread
From: Michael Ellerman @ 2019-02-19  1:02 UTC (permalink / raw)
  To: Meelis Roos, linuxppc-dev; +Cc: malat

Meelis Roos <mroos@linux.ee> writes:

>> Rather than relying on that we can pass an explict end_pos based on
>> the sizeof(vrsave). The result should be exactly the same but it's
>> more obviously not over-reading/writing the stack and it avoids the
>> compiler warning.
>
> It works on my PowerMac G4 with Debian-ports unstable with gcc 8.

Thanks.

cheers

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: powerpc/ptrace: Simplify vr_get/set() to avoid GCC warning
  2019-02-15  6:14 [PATCH] powerpc/ptrace: Simplify vr_get/set() to avoid GCC warning Michael Ellerman
  2019-02-15  8:09 ` Mathieu Malaterre
  2019-02-16 19:15 ` Meelis Roos
@ 2019-02-22  9:47 ` Michael Ellerman
  2 siblings, 0 replies; 6+ messages in thread
From: Michael Ellerman @ 2019-02-22  9:47 UTC (permalink / raw)
  To: Michael Ellerman, linuxppc-dev; +Cc: malat, mroos

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 2220 bytes --]

On Fri, 2019-02-15 at 06:14:00 UTC, Michael Ellerman wrote:
> GCC 8 warns about the logic in vr_get/set(), which with -Werror breaks
> the build:
> 
>   In function ‘user_regset_copyin’,
>       inlined from ‘vr_set’ at arch/powerpc/kernel/ptrace.c:628:9:
>   include/linux/regset.h:295:4: error: ‘memcpy’ offset [-527, -529] is
>   out of the bounds [0, 16] of object ‘vrsave’ with type ‘union
>   <anonymous>’ [-Werror=array-bounds]
>   arch/powerpc/kernel/ptrace.c: In function ‘vr_set’:
>   arch/powerpc/kernel/ptrace.c:623:5: note: ‘vrsave’ declared here
>      } vrsave;
> 
> This has been identified as a regression in GCC, see GCC bug 88273.
> 
> However we can avoid the warning and also simplify the logic and make
> it more robust.
> 
> Currently we pass -1 as end_pos to user_regset_copyout(). This says
> "copy up to the end of the regset".
> 
> The definition of the regset is:
> 	[REGSET_VMX] = {
> 		.core_note_type = NT_PPC_VMX, .n = 34,
> 		.size = sizeof(vector128), .align = sizeof(vector128),
> 		.active = vr_active, .get = vr_get, .set = vr_set
> 	},
> 
> The end is calculated as (n * size), ie. 34 * sizeof(vector128).
> 
> In vr_get/set() we pass start_pos as 33 * sizeof(vector128), meaning
> we can copy up to sizeof(vector128) into/out-of vrsave.
> 
> The on-stack vrsave is defined as:
>   union {
> 	  elf_vrreg_t reg;
> 	  u32 word;
>   } vrsave;
> 
> And elf_vrreg_t is:
>   typedef __vector128 elf_vrreg_t;
> 
> So there is no bug, but we rely on all those sizes lining up,
> otherwise we would have a kernel stack exposure/overwrite on our
> hands.
> 
> Rather than relying on that we can pass an explict end_pos based on
> the sizeof(vrsave). The result should be exactly the same but it's
> more obviously not over-reading/writing the stack and it avoids the
> compiler warning.
> 
> Reported-by: Meelis Roos <mroos@linux.ee>
> Reported-by: Mathieu Malaterre <malat@debian.org>
> Cc: stable@vger.kernel.org
> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
> Tested-by: Mathieu Malaterre <malat@debian.org>

Applied to powerpc next.

https://git.kernel.org/powerpc/c/ca6d5149d2ad0a8d2f9c28cbe3798022

cheers

^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2019-02-22 10:32 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-02-15  6:14 [PATCH] powerpc/ptrace: Simplify vr_get/set() to avoid GCC warning Michael Ellerman
2019-02-15  8:09 ` Mathieu Malaterre
2019-02-15  9:20   ` Michael Ellerman
2019-02-16 19:15 ` Meelis Roos
2019-02-19  1:02   ` Michael Ellerman
2019-02-22  9:47 ` Michael Ellerman

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).