Re: [PATCH] x86/asm: Don't use rbp as temp register in csum_partial_copy_generic()

From	Josh Poimboeuf <jpoimboe@redhat.com>
Newsgroups	linux.kernel
Subject	Re: [PATCH] x86/asm: Don't use rbp as temp register in csum_partial_copy_generic()
Date	2017-05-04 18:20 +0200
Message-ID	<tDrDY-5WC-9@gated-at.bofh.it> (permalink)
References	<tDqoy-4PF-15@gated-at.bofh.it> <tDruh-5So-5@gated-at.bofh.it>
Organization	linux.* mail to news gateway

Show all headers | View raw

On Thu, May 04, 2017 at 03:56:49PM +0000, David Laight wrote:
> From: Josh Poimboeuf
> > Sent: 04 May 2017 15:52
> > Andrey Konovalov reported the following warning while fuzzing the kernel
> > with syzkaller:
> > 
> >   WARNING: kernel stack regs at ffff8800686869f8 in a.out:4933 has bad 'bp' value c3fc855a10167ec0
> > 
> > The unwinder dump revealed that rbp had a bad value when an interrupt
> > occurred in csum_partial_copy_generic().
> > 
> > That function saves rbp on the stack and then overwrites it, using it as
> > a scratch register.  That's problematic because it breaks stack traces
> > if an interrupt occurs in the middle of the function.
> 
> Does gcc guarantee not to use bp as a scratch register in leaf functions?

At least in practice, gcc doesn't touch rbp in leaf functions.  (I don't
know about guarantees.)

-- 
Josh

Thread

[PATCH] x86/asm: Don't use rbp as temp register in csum_partial_copy_generic() Josh Poimboeuf <jpoimboe@redhat.com> - 2017-05-04 17:00 +0200
  RE: [PATCH] x86/asm: Don't use rbp as temp register in  csum_partial_copy_generic() David Laight <David.Laight@ACULAB.COM> - 2017-05-04 18:10 +0200
    Re: [PATCH] x86/asm: Don't use rbp as temp register in  csum_partial_copy_generic() Josh Poimboeuf <jpoimboe@redhat.com> - 2017-05-04 18:20 +0200
  [tip:x86/urgent] x86/asm: Don't use RBP as a temporary register in  csum_partial_copy_generic() tip-bot for Josh Poimboeuf <tipbot@zytor.com> - 2017-05-05 10:30 +0200

csiph-web