linux-kernel - Re: [PATCH V2 0/7] usb-misc: sisusbvga: cleanup and bug fix

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Fri, 15 Jan 2016 18:07:24 +0000
From:	Jason Cooper <jason@...edaemon.net>
To:	Joe Perches <joe@...ches.com>
Cc:	Peter Senna Tschudin <peter.senna@...labora.com>,
	thomas@...ischhofer.net, gregkh@...uxfoundation.org,
	linux-usb@...r.kernel.org, linux-kernel@...r.kernel.org,
	sergei.shtylyov@...entembedded.com,
	Masahiro Yamada <yamada.m@...panasonic.com>
Subject: Re: [PATCH V2 0/7] usb-misc: sisusbvga: cleanup and bug fix

Hey Joe, Peter,

On Fri, Jan 15, 2016 at 09:51:49AM -0800, Joe Perches wrote:
> (cc'ing Masahiro Yamada and Jason Cooper for objdiff)
> 
> On Fri, 2016-01-15 at 18:41 +0100, Peter Senna Tschudin wrote:
...
> As far as I know, gcc has never made guarantees about
> object output for the same input content.

True, we may want to reach out to the folks doing reproducible builds.
They may have seen/diagnosed/fixed this already.

> > # But here is the interesting part. Even compiling the exact same source code
> > # produces different results
> > $ git checkout -- .
> > $ md5sum drivers/usb/misc/sisusbvga/sisusb.c
> > d6ffbd44f3f1cf81fd55ce84441ab889  drivers/usb/misc/sisusbvga/sisusb.c
> > 
> > $ make -j4 drivers/usb/misc/sisusbvga/sisusb.o
> > $ objdump -D drivers/usb/misc/sisusbvga/sisusb.o| \
> >   sed "s/^[[:space:]]\+[0-9a-f]\+//" > /tmp/base_dump_again
> > 
> > $ diff /tmp/base_dump /tmp/base_dump_again 
> >  9135,9136c9135,9136
> >  < :	8e 4d 31             	mov    0x31(%rbp),%cs
> >  < :	46 00 00             	rex.RX add %r8b,(%rax)
> >  ---
> >  > :	de 10                	ficom  (%rax)
> >  > :	33 46 00             	xor    0x0(%rsi),%eax
> >  9139c9139
> >  < :	4b 00 00             	rex.WXB add %al,(%r8)
> >  ---
> >  > :	00 4b 00             	add    %cl,0x0(%rbx)

Well, this clearly shows that the whitespace change is a red herring.
I'm just guessing here, but could you try with -j1 to see if you can
still reproduce this inconsistency?

Cleaning the tree between builds may also narrow down the suspects.

thx,

Jason.