linux-kernel - Re: [RFC][PATCH] epoll: allow EPOLLWAKEUP flag if PM

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:	Mon, 11 Nov 2013 22:22:55 +0100
From:	"Rafael J. Wysocki" <rjw@...k.pl>
To:	Amit Pundir <amit.pundir@...aro.org>
Cc:	Arve Hjønnevåg <arve@...roid.com>,
	Alexander Viro <viro@...iv.linux.org.uk>,
	John Stultz <john.stultz@...aro.org>,
	linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org,
	Android Kernel Team <kernel-team@...roid.com>
Subject: Re: [RFC][PATCH] epoll: allow EPOLLWAKEUP flag if PM_SLEEP is enabled

On Monday, November 11, 2013 11:26:31 PM Amit Pundir wrote:
> I stumbled upon ENOMEM error from epoll_ctl() while bringing up Android-4.4 
> on a device that does not yet support PM_SLEEP.

Well, Android depends on PM_SLEEP which is a known fact.

> While looking into the problem, I found that ep_create_wakeup_source()
> reports ENOMEM if wakeup_source_register() returns NULL.
> ep_create_wakeup_source() assumes that NULL is only returned if we run
> into ENOMEM but NULL is also returned when CONFIG_PM_SLEEP is disabled.

So the error value should not be ENOMEM.

> If CONFIG_PM_SLEEP is disabled, stripping the EPOLLWAKEUP flag seems to
> be a reasonable solution here, allowing the call to succeed, while
> dropping the wakeup logic.  While returning EINVAL might also be a good
> solution, stripping the flag seems to follow the established behavior,
> as is done when the process doesn't have sufficient capabilities to
> block suspend.

Which is a different thing.

> I'd appreciate any thoughts or feedback!
>
> Signed-off-by: Amit Pundir <amit.pundir@...aro.org>
> ---
>  fs/eventpoll.c |    3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/eventpoll.c b/fs/eventpoll.c
> index 473e09d..7a83079 100644
> --- a/fs/eventpoll.c
> +++ b/fs/eventpoll.c
> @@ -1820,7 +1820,8 @@ SYSCALL_DEFINE4(epoll_ctl, int, epfd, int, op, int, fd,
>  		goto error_tgt_fput;
>  
>  	/* Check if EPOLLWAKEUP is allowed */
> -	if ((epds.events & EPOLLWAKEUP) && !capable(CAP_BLOCK_SUSPEND))
> +	if ((epds.events & EPOLLWAKEUP) &&
> +		(!capable(CAP_BLOCK_SUSPEND) || !IS_ENABLED(CONFIG_PM_SLEEP)))

I don't particularly like mixing the checking of CONFIG_ symbols with code like
this.

I'd create a static inline function taking a pointer to epds.events that would
do the

	if ((epds.events & EPOLLWAKEUP) && !capable(CAP_BLOCK_SUSPEND))
 		epds.events &= ~EPOLLWAKEUP;

for CONFIG_PM_SLEEP set and

  		epds.events &= ~EPOLLWAKEUP;

unconditionally for CONFIG_PM_SLEEP unset.

Thanks!

-- 
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.
--
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/