[PATCH] powerpc: 32 bit getcpu VDSO function uses 64 bit instructions

[PATCH] powerpc: 32 bit getcpu VDSO function uses 64 bit instructions

[Anton Blanchard]

I used some 64 bit instructions when adding the 32 bit getcpu VDSO
function. Fix it.

Fixes: 18ad51dd342a ("powerpc: Add VDSO version of getcpu")
Cc: stable@vger.kernel.org
Signed-off-by: Anton Blanchard <anton@samba.org>
---
 arch/powerpc/kernel/vdso32/getcpu.S | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/powerpc/kernel/vdso32/getcpu.S b/arch/powerpc/kernel/vdso32/getcpu.S
index 23eb9a9..c62be60 100644
--- a/arch/powerpc/kernel/vdso32/getcpu.S
+++ b/arch/powerpc/kernel/vdso32/getcpu.S
@@ -30,8 +30,8 @@
 V_FUNCTION_BEGIN(__kernel_getcpu)
   .cfi_startproc
 	mfspr	r5,SPRN_SPRG_VDSO_READ
-	cmpdi	cr0,r3,0
-	cmpdi	cr1,r4,0
+	cmpwi	cr0,r3,0
+	cmpwi	cr1,r4,0
 	clrlwi  r6,r5,16
 	rlwinm  r7,r5,16,31-15,31-0
 	beq	cr0,1f
-- 
2.1.0

Re: [PATCH] powerpc: 32 bit getcpu VDSO function uses 64 bit instructions

[Michael Ellerman]

On Thu, 2014-11-27 at 08:11 +1100, Anton Blanchard wrote:
> I used some 64 bit instructions when adding the 32 bit getcpu VDSO
> function. Fix it.

Ouch. The symptom is a SIGILL I presume?

Could we catch this by forcing -m32 in the CFLAGS for vdso32 ?

cheers

Re: [PATCH] powerpc: 32 bit getcpu VDSO function uses 64 bit instructions

[Segher Boessenkool]

On Thu, Nov 27, 2014 at 09:38:17AM +1100, Michael Ellerman wrote:
> On Thu, 2014-11-27 at 08:11 +1100, Anton Blanchard wrote:
> > I used some 64 bit instructions when adding the 32 bit getcpu VDSO
> > function. Fix it.
> 
> Ouch. The symptom is a SIGILL I presume?
> 
> Could we catch this by forcing -m32 in the CFLAGS for vdso32 ?

GCC has added -many to the assembler flags for over ten years now, so
no that will not work.  You can use -mppc or similar with the assembler
if you invoke it correctly (use  $(CC) -print-prog-name=as  to figure
out how to call the assembler, if you want to stay sane and use the
same one as the rest of the toolchain you use).


Segher

Re: [PATCH] powerpc: 32 bit getcpu VDSO function uses 64 bit instructions

[Segher Boessenkool]

On Wed, Nov 26, 2014 at 05:23:18PM -0600, Segher Boessenkool wrote:
> GCC has added -many to the assembler flags for over ten years now, so
> no that will not work.  You can use -mppc or similar with the assembler
> if you invoke it correctly (use  $(CC) -print-prog-name=as  to figure
s/correctly/directly/
> out how to call the assembler, if you want to stay sane and use the
> same one as the rest of the toolchain you use).
> 
> 
> Segher

Re: [PATCH] powerpc: 32 bit getcpu VDSO function uses 64 bit instructions

[Peter Bergner]

On Thu, 2014-11-27 at 09:38 +1100, Michael Ellerman wrote:
> On Thu, 2014-11-27 at 08:11 +1100, Anton Blanchard wrote:
> > I used some 64 bit instructions when adding the 32 bit getcpu VDSO
> > function. Fix it.
> 
> Ouch. The symptom is a SIGILL I presume?

Nope, you don't get a SIGILL when executing 64-bit instructions in
32-bit mode, so it'll happily just execute the instruction, doing
a full 64-bit compare.  I'm guessing that the upper 32-bits of both
r3 and r4 contain zeros, so we're probably just getting lucky.


> Could we catch this by forcing -m32 in the CFLAGS for vdso32 ?

As Segher mentioned, GCC passing -many down to the assembler means
-m32 won't help.  It was due to Anton disabling that gcc "feature",
that this was caught.

Peter

Re: [PATCH] powerpc: 32 bit getcpu VDSO function uses 64 bit instructions

[Segher Boessenkool]

On Wed, Nov 26, 2014 at 05:50:27PM -0600, Peter Bergner wrote:
> On Thu, 2014-11-27 at 09:38 +1100, Michael Ellerman wrote:
> > On Thu, 2014-11-27 at 08:11 +1100, Anton Blanchard wrote:
> > > I used some 64 bit instructions when adding the 32 bit getcpu VDSO
> > > function. Fix it.
> > 
> > Ouch. The symptom is a SIGILL I presume?
> 
> Nope, you don't get a SIGILL when executing 64-bit instructions in
> 32-bit mode, so it'll happily just execute the instruction, doing
> a full 64-bit compare.  I'm guessing that the upper 32-bits of both
> r3 and r4 contain zeros, so we're probably just getting lucky.

You will get a SIGILL if you run on 32-bit hardware.

> > Could we catch this by forcing -m32 in the CFLAGS for vdso32 ?
> 
> As Segher mentioned, GCC passing -many down to the assembler means
> -m32 won't help.  It was due to Anton disabling that gcc "feature",
> that this was caught.

There are extremely many complex failure cases without the -many.
Feel free to work on removing it, I won't :-)


Segher

Re: [PATCH] powerpc: 32 bit getcpu VDSO function uses 64 bit instructions

[Peter Bergner]

On Thu, 2014-11-27 at 10:08 -0600, Segher Boessenkool wrote:
> On Wed, Nov 26, 2014 at 05:50:27PM -0600, Peter Bergner wrote:
> > Nope, you don't get a SIGILL when executing 64-bit instructions in
> > 32-bit mode, so it'll happily just execute the instruction, doing
> > a full 64-bit compare.  I'm guessing that the upper 32-bits of both
> > r3 and r4 contain zeros, so we're probably just getting lucky.
> 
> You will get a SIGILL if you run on 32-bit hardware.

Ha, I completely forgot about 32-bit hardware.  Anyway, I looked
at the ISA, and cmpdi and cmpwi are just extended mnemonics for
cmpi, with cmpdi setting the L field to 1.  Probably on 32-bit
hardware, the hardware is just ignoring the L bit being set and
doing a cmpwi for us???

Peter

Re: [PATCH] powerpc: 32 bit getcpu VDSO function uses 64 bit instructions

[Andreas Schwab]

Segher Boessenkool <segher@kernel.crashing.org> writes:

> On Wed, Nov 26, 2014 at 05:50:27PM -0600, Peter Bergner wrote:
>> On Thu, 2014-11-27 at 09:38 +1100, Michael Ellerman wrote:
>> > On Thu, 2014-11-27 at 08:11 +1100, Anton Blanchard wrote:
>> > > I used some 64 bit instructions when adding the 32 bit getcpu VDSO
>> > > function. Fix it.
>> > 
>> > Ouch. The symptom is a SIGILL I presume?
>> 
>> Nope, you don't get a SIGILL when executing 64-bit instructions in
>> 32-bit mode, so it'll happily just execute the instruction, doing
>> a full 64-bit compare.  I'm guessing that the upper 32-bits of both
>> r3 and r4 contain zeros, so we're probably just getting lucky.
>
> You will get a SIGILL if you run on 32-bit hardware.

Not on the 7447A, fwiw.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: [PATCH] powerpc: 32 bit getcpu VDSO function uses 64 bit instructions

[Segher Boessenkool]

On Thu, Nov 27, 2014 at 11:41:40AM -0600, Peter Bergner wrote:
> On Thu, 2014-11-27 at 10:08 -0600, Segher Boessenkool wrote:
> > On Wed, Nov 26, 2014 at 05:50:27PM -0600, Peter Bergner wrote:
> > > Nope, you don't get a SIGILL when executing 64-bit instructions in
> > > 32-bit mode, so it'll happily just execute the instruction, doing
> > > a full 64-bit compare.  I'm guessing that the upper 32-bits of both
> > > r3 and r4 contain zeros, so we're probably just getting lucky.
> > 
> > You will get a SIGILL if you run on 32-bit hardware.
> 
> Ha, I completely forgot about 32-bit hardware.  Anyway, I looked
> at the ISA, and cmpdi and cmpwi are just extended mnemonics for
> cmpi, with cmpdi setting the L field to 1.  Probably on 32-bit
> hardware, the hardware is just ignoring the L bit being set and
> doing a cmpwi for us???

Huh.  Yes, maybe some implementations do that.

The good news is that those then compute the correct thing ;-)

Can QEMU help catch such bugs more reliably?


Segher

Re: [PATCH] powerpc: 32 bit getcpu VDSO function uses 64 bit instructions

[Benjamin Herrenschmidt]

On Thu, 2014-11-27 at 14:50 -0600, Segher Boessenkool wrote:
> On Thu, Nov 27, 2014 at 11:41:40AM -0600, Peter Bergner wrote:
> > On Thu, 2014-11-27 at 10:08 -0600, Segher Boessenkool wrote:
> > > On Wed, Nov 26, 2014 at 05:50:27PM -0600, Peter Bergner wrote:
> > > > Nope, you don't get a SIGILL when executing 64-bit instructions in
> > > > 32-bit mode, so it'll happily just execute the instruction, doing
> > > > a full 64-bit compare.  I'm guessing that the upper 32-bits of both
> > > > r3 and r4 contain zeros, so we're probably just getting lucky.
> > > 
> > > You will get a SIGILL if you run on 32-bit hardware.
> > 
> > Ha, I completely forgot about 32-bit hardware.  Anyway, I looked
> > at the ISA, and cmpdi and cmpwi are just extended mnemonics for
> > cmpi, with cmpdi setting the L field to 1.  Probably on 32-bit
> > hardware, the hardware is just ignoring the L bit being set and
> > doing a cmpwi for us???
> 
> Huh.  Yes, maybe some implementations do that.
> 
> The good news is that those then compute the correct thing ;-)
> 
> Can QEMU help catch such bugs more reliably?

That's all moot, that piece of code only exist on 64-bit kernels :-)

So the only risk here is the very remote and unlikely case where the
register might contain 0 in the low 32-bits and some garbage in the top.

Cheers,
Ben.