linux-kernel - Re: sigaltstack breaks swapcontext()

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: <568EB81D.9030209@list.ru>
Date:	Thu, 7 Jan 2016 22:10:21 +0300
From:	Stas Sergeev <stsp@...t.ru>
To:	Andy Lutomirski <luto@...capital.net>
Cc:	Linux kernel <linux-kernel@...r.kernel.org>
Subject: Re: sigaltstack breaks swapcontext()

07.01.2016 20:23, Andy Lutomirski пишет:
> On Thu, Jan 7, 2016 at 7:33 AM, Stas Sergeev <stsp@...t.ru> wrote:
>> 06.01.2016 22:53, Andy Lutomirski пишет:
>>> On Wed, Jan 6, 2016 at 11:31 AM, Stas Sergeev <stsp@...t.ru> wrote:
>>>> Exactly.
>>>> Do you think this can be ignored?
>>>> A man page should then be corrected with EPERM and the
>>>> above note removed, right?
>>>>
>>> I think it can be ignored.  I'd go the SS_FORCE route, though, to
>>> maintain POSIX compliance.
>> I think such a flag would be a wrong thing to do.
>> Allowing only SS_DISABLE (without any new flags) keeps
>> you still "compatible with posix", and anything beyond
>> SS_DISABLE in a sighandler is not needed.
>>
>> So I think we only have the following options:
>> 1. Remove the check and forget (if anything, glibc can
>> add the EPERM check to stay compatible with crap).
>> 2. Allow only SS_DISABLE. This will mean a large patch,
>> touching all arches, but the bonus is the compatibility
>> with posix, that no one needs in this particular case.
> Why does allowing SS_DISABLE require touching all arches?
I mean, if we also consistently return SA_ONSTACK even
after SS_DISABLE. This will require kernel to save the
ss_flags separately, and not to use "if (ss_size)" checks.
Of course if we don't want to return SA_ONSTACK, we
should just remove EPERM as I don't think it serves any
other purpose than that.