linux-kernel - Re: [PATCH FIX] traps: x86: correct copy/paste bug: a trap is a GATE

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: Windows password security audit tool. GUI, reports in PDF.

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

Date:	Sat, 4 Oct 2008 14:13:33 +0200
From:	Ingo Molnar <mingo@...e.hu>
To:	Alexander van Heukelum <heukelum@...lshack.com>
Cc:	LKML <linux-kernel@...r.kernel.org>,
	Thomas Gleixner <tglx@...utronix.de>,
	"H. Peter Anvin" <hpa@...or.com>, Yinghai Lu <yinghai@...nel.org>,
	Jeremy Fitzhardinge <jeremy@...p.org>
Subject: Re: [PATCH FIX] traps: x86: correct copy/paste bug: a trap is a
	GATE_TRAP


* Alexander van Heukelum <heukelum@...lshack.com> wrote:

> > > > This series unifies traps_32.c and traps_64.c.
> > > 
> > > wow, very nice! I've applied them to tip/x86/traps:
> > > 
> > > f58f3d5: traps: x86: finalize unification of traps.c
> > > bf395d6: traps: x86: make traps_32.c and traps_64.c equal
> > > f156f35: traps: x86: various noop-changes preparing for unification of traps_xx.c
> > > 70cfe30: traps: x86_64: use task_pid_nr(tsk) instead of tsk->pid in do_general_protection
> > > dc89ce0: traps: i386: expand clear_mem_error and remove from mach_traps.h
> > > 6f8063f: traps: x86_64: make io_check_error equal to the one on i386
> > > d025445: traps: i386: use preempt_conditional_sti/cli in do_int3
> > > 2180afa: traps: x86_64: make math_state_restore more like i386
> > > 686cc4a: traps: x86: converge trap_init functions
> > 
> > -tip testing found a spontaneus reboot bug on two 32-bit systems (one 
> > Intel and one AMD testbox), and i've bisected it down to:
> > 
> > | 686cc4a0c1ca92bffbc22a897c3b433dadbbf444 is first bad commit
> > | commit 686cc4a0c1ca92bffbc22a897c3b433dadbbf444
> > | Author: Alexander van Heukelum <heukelum@...tmail.fm>
> > | Date:   Fri Oct 3 22:00:32 2008 +0200
> > |
> > |     traps: x86: converge trap_init functions
> > 
> > config attached. The bisection log:
> > 
> >  # bad:  [a229a9da] Merge branch 'timers/urgent'
> >  # good: [27de5e39] Merge branch 'out-of-tree'
> >  # good: [8e9cb9db] Merge branch 'tracing/ring-buffer'
> >  # bad:  [70cfe30f] traps: x86_64: use task_pid_nr(tsk) instead of tsk
> >  # bad:  [d0254456] traps: i386: use preempt_conditional_sti/cli in do
> >  # bad:  [2180afaf] traps: x86_64: make math_state_restore more like i
> >  # bad:  [686cc4a0] traps: x86: converge trap_init functions
> > 
> > so i've excluded these commits from tip/master for now.
> > 
> > there's no serial log entry visible - the spontaneous reboot happens at 
> > around when we hit user-space.
> > 
> > I suspect syscall entry setup might be borked - the stack frames of call 
> > gates versus interrupt gates are different and it's easy to make a small 
> > mistake there with such effects.
> > 
> > 	Ingo
> 
> *blush*
> 
> You were so right. I have no idea how this has slipped
> through testing. Could you see if this on top of the
> traps branch makes things go again? I'll get you some
> replacement patches to make the whole thing bisectable
> again.

>  static inline void set_system_trap_gate(unsigned int n, void *addr)
>  {
>  	BUG_ON((unsigned)n > 0xFF);
> -	_set_gate(n, GATE_INTERRUPT, addr, 0x3, 0, __KERNEL_CS);
> +	_set_gate(n, GATE_TRAP, addr, 0x3, 0, __KERNEL_CS);
>  }

indeed, that would explain it :)

i've added your delta fix, and will propagate/squash it back to 1/9 if 
it tests out fine. (it's worth having this series fully bisectable)

	Ingo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/