linux-kernel - Re: ->poll() instances shouldn't be indefinitely blocking

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]

Date:	Fri, 27 Nov 2015 09:49:11 -0800
From:	Linus Torvalds <torvalds@...ux-foundation.org>
To:	Mauro Carvalho Chehab <mchehab@....samsung.com>
Cc:	Al Viro <viro@...iv.linux.org.uk>,
	Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
	linux-fsdevel <linux-fsdevel@...r.kernel.org>,
	Linux Media Mailing List <linux-media@...r.kernel.org>,
	Kukjin Kim <kgene@...nel.org>,
	Krzysztof Kozlowski <k.kozlowski@...sung.com>,
	Hans Verkuil <hverkuil@...all.nl>
Subject: Re: ->poll() instances shouldn't be indefinitely blocking

On Fri, Nov 27, 2015 at 7:18 AM, Mauro Carvalho Chehab
<mchehab@....samsung.com> wrote:
> Al Viro <viro@...IV.linux.org.uk> escreveu:
>
>> Take a look at this:
>> static unsigned int gsc_m2m_poll(struct file *file,
>>                                         struct poll_table_struct *wait)
>> {
>>         struct gsc_ctx *ctx = fh_to_ctx(file->private_data);
>>         struct gsc_dev *gsc = ctx->gsc_dev;
>>         int ret;
>>
>>         if (mutex_lock_interruptible(&gsc->lock))
>>                 return -ERESTARTSYS;
>>
>>         ret = v4l2_m2m_poll(file, ctx->m2m_ctx, wait);
>>         mutex_unlock(&gsc->lock);
>>
>>         return ret;
>> }
>>
>> a) ->poll() should not return -E...; callers expect just a bitmap of
>> POLL... values.
>
> Yeah. We fixed issues like that on other drivers along the time. I guess
> this is a some bad code that people just cut-and-paste from legacy drivers
> without looking into it.

Actually, while returning -ERESTARTSYS is bogus, returning _zero_
would not be. The top-level poll() code will happily notice the
signal, and return -EINTR like poll should (unless something else is
pending, in which case it will return zero and the bits set for that
something else).

So having a driver with a ->poll() function that does that kind of
conditional locking is not wrong per se. It's just he return value
that is crap.

I also do wonder if we might not make the generic code a bit more
robust wrt things like this. The bitmask we use is only about the low
bits, so we *could* certainly allow the driver poll() functions to
return errors - possibly just ignoring them. Or perhaps have a
WARN_ON_OCNE() to find them.

Al, what do you think? The whole "generic code should be robust wrt
drivers making silly mistakes" just sounds like a good idea. Finding
these things through code inspection is all well and good, but having
a nice warning report from users might be even better.

                Linus
--
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/