All of lore.kernel.org
 help / color / mirror / Atom feed
From: Arnd Bergmann <arnd@arndb.de>
To: Hans Verkuil <hverkuil@xs4all.nl>
Cc: linux-samsung-soc@vger.kernel.org,
	Mauro Carvalho Chehab <mchehab@osg.samsung.com>,
	y2038@lists.linaro.org, linux-api@vger.kernel.org,
	linux-kernel@vger.kernel.org, linux-media@vger.kernel.org
Subject: Re: [PATCH v2 7/9] [media] v4l2: introduce v4l2_timeval
Date: Fri, 18 Sep 2015 11:43:24 +0200	[thread overview]
Message-ID: <9019880.VVdOR6WRt1@wuerfel> (raw)
In-Reply-To: <55FBD90C.3020301@xs4all.nl>

On Friday 18 September 2015 11:27:40 Hans Verkuil wrote:
> Ah, OK. Got it.
> 
> I think this is dependent on the upcoming media workshop next month. If we
> decide to redesign v4l2_buffer anyway, then we can avoid timeval completely.
> And the only place where we would need to convert it in the compat code
> hidden in the v4l2 core (likely v4l2-ioctl.c).

Ah, I think I understood the idea now, I missed that earlier when you mention
the idea.

So what you are saying here is that you could come up with a new unambiguous
(using only __u32 and __u64 types and no pointers) format that gets exposed
to a new set of ioctls, and then change the handling of the existing three
formats (native 64-bit, traditional 32-bit, and 32-bit with 64-bit time_t)
so they get converted into the new format by the common ioctl handling code?

> I am not really keen on having v4l2_timeval in all these drivers. I would
> have to check them anyway since I suspect that in several drivers the local
> timeval variable can be avoided by rewriting that part of the driver.

I've tried to do that for all the drivers where I could find an easy solution
in patch 6/9, but I'm sure you can do it for a couple more.

We could also do a lightweight redesign and use 'timespec64' internally
in all the drivers and then convert that to 'timeval' or the 64-bit
format of that in the ioctl handler. This is also something I tried at
some point but then found it a bit more intuitive to leave the normal ioctl
path alone and have an explicit type.

> Personally I am in favor of a redesigned v4l2_buffer: it's awkward to use
> with multiplanar formats, there is cruft in there that can be removed (timecode),
> and there is little space for additions (HW-specific timecodes, more buffer
> meta data, etc).
> 
> We'll see.

Ok.

	Arnd
_______________________________________________
Y2038 mailing list
Y2038@lists.linaro.org
https://lists.linaro.org/mailman/listinfo/y2038

WARNING: multiple messages have this Message-ID (diff)
From: Arnd Bergmann <arnd@arndb.de>
To: Hans Verkuil <hverkuil@xs4all.nl>
Cc: linux-media@vger.kernel.org, linux-kernel@vger.kernel.org,
	y2038@lists.linaro.org,
	Mauro Carvalho Chehab <mchehab@osg.samsung.com>,
	linux-api@vger.kernel.org, linux-samsung-soc@vger.kernel.org
Subject: Re: [PATCH v2 7/9] [media] v4l2: introduce v4l2_timeval
Date: Fri, 18 Sep 2015 11:43:24 +0200	[thread overview]
Message-ID: <9019880.VVdOR6WRt1@wuerfel> (raw)
In-Reply-To: <55FBD90C.3020301@xs4all.nl>

On Friday 18 September 2015 11:27:40 Hans Verkuil wrote:
> Ah, OK. Got it.
> 
> I think this is dependent on the upcoming media workshop next month. If we
> decide to redesign v4l2_buffer anyway, then we can avoid timeval completely.
> And the only place where we would need to convert it in the compat code
> hidden in the v4l2 core (likely v4l2-ioctl.c).

Ah, I think I understood the idea now, I missed that earlier when you mention
the idea.

So what you are saying here is that you could come up with a new unambiguous
(using only __u32 and __u64 types and no pointers) format that gets exposed
to a new set of ioctls, and then change the handling of the existing three
formats (native 64-bit, traditional 32-bit, and 32-bit with 64-bit time_t)
so they get converted into the new format by the common ioctl handling code?

> I am not really keen on having v4l2_timeval in all these drivers. I would
> have to check them anyway since I suspect that in several drivers the local
> timeval variable can be avoided by rewriting that part of the driver.

I've tried to do that for all the drivers where I could find an easy solution
in patch 6/9, but I'm sure you can do it for a couple more.

We could also do a lightweight redesign and use 'timespec64' internally
in all the drivers and then convert that to 'timeval' or the 64-bit
format of that in the ioctl handler. This is also something I tried at
some point but then found it a bit more intuitive to leave the normal ioctl
path alone and have an explicit type.

> Personally I am in favor of a redesigned v4l2_buffer: it's awkward to use
> with multiplanar formats, there is cruft in there that can be removed (timecode),
> and there is little space for additions (HW-specific timecodes, more buffer
> meta data, etc).
> 
> We'll see.

Ok.

	Arnd

  reply	other threads:[~2015-09-18  9:43 UTC|newest]

Thread overview: 40+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-09-17 21:19 [PATCH v2 0/9] [media] y2038 conversion for subsystem Arnd Bergmann
2015-09-17 21:19 ` [PATCH v2 1/9] [media] dvb: use ktime_t for internal timeout Arnd Bergmann
2015-09-17 21:19   ` Arnd Bergmann
2015-09-17 21:19 ` [PATCH v2 2/9] [media] dvb: remove unused systime() function Arnd Bergmann
2015-09-17 21:19   ` Arnd Bergmann
2015-09-17 21:19 ` [PATCH v2 3/9] [media] dvb: don't use 'time_t' in event ioctl Arnd Bergmann
2015-09-17 21:19 ` [PATCH v2 4/9] [media] exynos4-is: use monotonic timestamps as advertized Arnd Bergmann
2015-09-17 21:19 ` [PATCH v2 5/9] [media] make VIDIOC_DQEVENT work with 64-bit time_t Arnd Bergmann
2015-09-17 21:19   ` Arnd Bergmann
2015-09-17 21:19 ` [PATCH v2 6/9] [media] use v4l2_get_timestamp where possible Arnd Bergmann
2015-09-17 21:19   ` Arnd Bergmann
2015-09-17 21:19 ` [PATCH v2 7/9] [media] v4l2: introduce v4l2_timeval Arnd Bergmann
2015-09-17 21:19   ` Arnd Bergmann
     [not found]   ` <1442524780-781677-8-git-send-email-arnd-r2nGTMty4D4@public.gmane.org>
2015-09-18  8:05     ` Hans Verkuil
2015-09-18  8:05       ` Hans Verkuil
2015-09-18  9:09       ` Arnd Bergmann
2015-09-18  9:27         ` Hans Verkuil
2015-09-18  9:43           ` Arnd Bergmann [this message]
2015-09-18  9:43             ` Arnd Bergmann
2015-09-18  9:52             ` Hans Verkuil
2015-09-18  9:52               ` Hans Verkuil
2015-09-18 10:08               ` Arnd Bergmann
2015-09-18 10:22                 ` Hans Verkuil
2015-09-18 10:22                   ` Hans Verkuil
     [not found] ` <1442524780-781677-1-git-send-email-arnd-r2nGTMty4D4@public.gmane.org>
2015-09-17 21:19   ` [PATCH v2 8/9] [media] handle 64-bit time_t in v4l2_buffer Arnd Bergmann
2015-09-17 21:19     ` Arnd Bergmann
2015-09-18  7:18     ` Hans Verkuil
2015-09-18  7:18       ` Hans Verkuil
2015-09-18  9:26       ` Arnd Bergmann
2015-09-18  9:49         ` Hans Verkuil
2015-09-18 10:14           ` Arnd Bergmann
2015-09-18 10:14             ` Arnd Bergmann
2015-09-17 21:19 ` [PATCH v2 9/9] [media] omap3isp: support 64-bit version of omap3isp_stat_data Arnd Bergmann
2015-09-17 21:19   ` Arnd Bergmann
     [not found]   ` <1442524780-781677-10-git-send-email-arnd-r2nGTMty4D4@public.gmane.org>
2015-11-09 20:09     ` Laurent Pinchart
2015-11-09 20:09       ` Laurent Pinchart
2015-11-09 20:30       ` Arnd Bergmann
2015-11-09 20:30         ` [Y2038] " Arnd Bergmann
2015-11-09 21:04         ` Laurent Pinchart
2015-11-09 21:04           ` Laurent Pinchart

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=9019880.VVdOR6WRt1@wuerfel \
    --to=arnd@arndb.de \
    --cc=hverkuil@xs4all.nl \
    --cc=linux-api@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-media@vger.kernel.org \
    --cc=linux-samsung-soc@vger.kernel.org \
    --cc=mchehab@osg.samsung.com \
    --cc=y2038@lists.linaro.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body. 
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.