linux-kernel - Re: [RFC] TIF_NOTIFY_RESUME, arch/*/*/*signal*.c and all such

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

Message-ID: <20120429041205.GY6871@ZenIV.linux.org.uk>
Date:	Sun, 29 Apr 2012 05:12:05 +0100
From:	Al Viro <viro@...IV.linux.org.uk>
To:	Martin Schwidefsky <schwidefsky@...ibm.com>
Cc:	Oleg Nesterov <oleg@...hat.com>,
	Linus Torvalds <torvalds@...ux-foundation.org>,
	linux-arch@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [RFC] TIF_NOTIFY_RESUME, arch/*/*/*signal*.c and all such

On Sat, Apr 28, 2012 at 12:32:35AM +0100, Al Viro wrote:
> On Sat, Apr 28, 2012 at 12:15:26AM +0100, Al Viro wrote:
> 
> > I think all such architectures need that check lifted to do_notify_resume()
> > (and the rest needs it killed, of course).  Including x86, by the look
> > of it - we _probably_ can't get there with TIF_NOTIFY_RESUME and
> > !user_mode(regs), but I'm not entirely sure of that.  arm is in about the
> > same situation; alpha, ppc{32,64}, sparc{32,64} and m68k really can't get
> > there like that (they all check it in the asm glue).  mips probably might,
> > unless I'm misreading their ret_from_fork()...  Fun.
> 
> 	Speaking of user_mode() oddities - may I politely inquire what had
> been smoked to inspire this (in arch/s390/kernel/signal.c):
>         /* This is the legacy signal stack switching. */
>         else if (!user_mode(regs) &&
>                  !(ka->sa.sa_flags & SA_RESTORER) &&
>                  ka->sa.sa_restorer) {
>                 sp = (unsigned long) ka->sa.sa_restorer;
>         }
> especially when all paths leading to that come through do_signal() that does
>         if (!user_mode(regs))
>                 return;
> on the same regs.  It had been like that since 2.3.99pre8 when s390 went
> into the tree...  It looks vaguely similar to i386
>                         /* This is the legacy signal stack switching. */
>                         if ((regs->ss & 0xffff) != __USER_DS &&
>                                 !(ka->sa.sa_flags & SA_RESTORER) &&
>                                         ka->sa.sa_restorer)
>                                 sp = (unsigned long) ka->sa.sa_restorer;
> but there the code is at least not unreachable...

While we are at it, can we *ever* reach do_signal() (nevermind deep in its
guts) with !user_mode(regs)?
AFAICS, for 31bit possible paths are:
do_signal()
	<- sysc_sigpending
		<- sysc_work
			<- sysc_tif, having checked for user_mode(%r11)
	<- io_sigpending
		<- io_work_tif
			<- io_work_user
				<- io_work, having checked for user_mode(%r11)

and identical for 64bit.  *All* paths into do_signal() go through
        tm      __PT_PSW+1(%r11),0x01   # returning to user ?
and proceed towards do_signal() only if the bit is set.  Which is precisely
what user_mode(%r11) is...

What the hell is going on in that code?
--
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/