From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: Vincent MAILHOL <mailhol.vincent@wanadoo.fr>
Cc: David Laight <David.Laight@aculab.com>,
Alan Stern <stern@rowland.harvard.edu>,
Marc Kleine-Budde <mkl@pengutronix.de>,
Rhett Aultman <rhett.aultman@samsara.com>,
"linux-usb@vger.kernel.org" <linux-usb@vger.kernel.org>,
linux-can <linux-can@vger.kernel.org>,
Oliver Neukum <oneukum@suse.com>
Subject: Re: [PATCH v3 1/2] drivers: usb/core/urb: Add URB_FREE_COHERENT
Date: Wed, 22 Jun 2022 14:23:46 +0200 [thread overview]
Message-ID: <YrMJ0vXR84cISTse@kroah.com> (raw)
In-Reply-To: <CAMZ6Rq+e+i0xGuWeMaGyTVyLJy=q2vQZqXjYoACws=T_B-FOGQ@mail.gmail.com>
On Wed, Jun 22, 2022 at 07:34:57PM +0900, Vincent MAILHOL wrote:
> On Wed. 22 Jun 2022 at 18:44, Greg Kroah-Hartman
> <gregkh@linuxfoundation.org> wrote:
> > On Wed, Jun 22, 2022 at 09:22:12AM +0000, David Laight wrote:
> > > From: Vincent MAILHOL
> > > > Sent: 21 June 2022 16:56
> > > >
> > > > On Wed. 22 Jun 2022 at 00:13, Alan Stern <stern@rowland.harvard.edu> wrote:
> > > > > On Tue, Jun 21, 2022 at 11:59:16PM +0900, Vincent MAILHOL wrote:
> > > > > > I (probably wrongly) assumed that urb::transfer_buffer_length was the
> > > > > > allocated length and urb::actual_length was what was actually being
> > > > > > transferred. Right now, I am just confused. Seems that I need to study
> > > > > > a bit more and understand the real purpose of
> > > > > > urb::transfer_buffer_length because I still fail to understand in
> > > > > > which situation this can be different from the allocated length.
> > > > >
> > > > > urb->transfer_buffer_length is the amount of data that the driver wants
> > > > > to send or expects to receive. urb->actual_length is the amount of data
> > > > > that was actually sent or actually received.
> > > > >
> > > > > Neither of these values has to be the same as the size of the buffer --
> > > > > but they better not be bigger!
> > > >
> > > > Thanks. Now things are a bit clearer.
> > > > I guess that for the outcoming URB what I proposed made no sense. For
> > > > incoming URB, I guess that most of the drivers want to set
> > > > urb::transfer_buffer once for all with the allocated size and never
> > > > touch it again.
> > > >
> > > > Maybe the patch only makes sense of the incoming URB. Would it make
> > > > sense to keep it but with an additional check to trigger a dmesg
> > > > warning if this is used on an outcoming endpoint and with additional
> > > > comment that the URB_FREE_COHERENT requires urb::transfer_buffer to be
> > > > the allocated size?
> > >
> > > IIRC urb are pretty big.
> >
> > What exactly do you mean by "pretty big" here? And what is wrong with
> > that, I have never seen any issues with the current size of that
> > structure in any benchmark or performance results. All USB bottlenecks
> > that I know of are either in the hardware layer, or in the protocol
> > layer itself (i.e. usb-storage protocol).
> >
> > > You'd be unlucky if adding an extra field to hold the allocated
> > > size would ever need more memory.
> > > So it might just be worth saving the allocated size.
> >
> > Maybe, yes, then we could transition to allocating the urb and buffer at
> > the same time like we do partially for iso streams in an urb. But that
> > still might be overkill for just this one driver.
>
> Well, I wouldn't have proposed the patch if it only applied to a
> single driver. If we add a urb::allocated_transfer_size as suggested
> by David, I believe that the majority of the drivers using DMA memory
> will be able to rely on that URB_FREE_COHERENT flag for the garbage
> collection.
>
> The caveat, as you pointed before, is that the developper still needs
> to be aware of the limitations of DMA and that it should not be freed
> in an IRQ context. e.g. no call to usb_kill_anchored_urbs() or other
> functions that would lead to urb_destroy().
>
> > I'm curious as to why
> > a slow and tiny protocol like CAN needs to use usb_alloc_coherent() for
> > its buffers in the first place.
>
> The CAN protocol, in its latest revision, allows for transfer speed up
> to ~5Mbits. For low performance CPUs, this starts to be a significant
> load. Also, the CAN PDU being small (0 to 64 bytes), many small
> transfers occur.
And is the memcpy the actual issue here? Even tiny cpus can do large
and small memcopy very very very fast.
> Unfortunately I did not do any benchmark myself so I won't be able to
> back my explanation with numbers.
That might be the simplest solution here :)
thanks,
greg k-h
next prev parent reply other threads:[~2022-06-22 12:24 UTC|newest]
Thread overview: 72+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-06-03 19:52 [PATCH 001/001] can: gs_usb: gs_usb_open/close( ) fix memory leak Rhett Aultman
2022-06-04 2:11 ` [PATCH] " Vincent Mailhol
2022-06-04 2:26 ` Vincent MAILHOL
2022-06-04 14:08 ` Rhett Aultman
2022-06-04 14:41 ` [RFC PATCH] USB: core: urb: add new transfer flag URB_FREE_COHERENT Vincent Mailhol
2022-06-04 16:40 ` Alan Stern
2022-06-05 2:04 ` Vincent MAILHOL
2022-06-05 6:00 ` Oliver Neukum
2022-06-05 13:45 ` Vincent MAILHOL
2022-06-07 9:49 ` Oliver Neukum
2022-06-07 10:18 ` Vincent MAILHOL
2022-06-07 11:46 ` Oliver Neukum
2022-06-07 12:12 ` Vincent MAILHOL
2022-06-05 2:15 ` [RFC PATCH v2] usb: " Vincent Mailhol
2022-06-04 14:53 ` [PATCH] can: gs_usb: gs_usb_open/close( ) fix memory leak Vincent MAILHOL
2022-06-09 20:47 ` [PATCH v2 0/3] URB_FREE_COHERENT gs_usb memory leak fix Rhett Aultman
2022-06-09 20:47 ` [PATCH v2 1/3] drivers: usb/core/urb: Add URB_FREE_COHERENT Rhett Aultman
2022-06-10 0:18 ` Vincent MAILHOL
2022-06-10 10:46 ` Marc Kleine-Budde
2022-06-09 20:47 ` [PATCH v2 2/3] drivers: usb/core/urb: allow URB_FREE_COHERENT Rhett Aultman
2022-06-09 23:18 ` Vincent Mailhol
2022-06-09 20:47 ` [PATCH v2 3/3] can: gs_usb: fix DMA memory leak on close Rhett Aultman
2022-06-10 0:05 ` Vincent Mailhol
2022-06-10 1:28 ` Vincent Mailhol
2022-06-10 21:33 ` [PATCH v3 0/2] URB_FREE_COHERENT gs_usb memory leak fix Rhett Aultman
2022-06-10 21:33 ` [PATCH v3 1/2] drivers: usb/core/urb: Add URB_FREE_COHERENT Rhett Aultman
2022-06-11 15:31 ` Marc Kleine-Budde
2022-06-11 16:06 ` Vincent MAILHOL
2022-06-21 13:51 ` Greg Kroah-Hartman
2022-06-21 14:59 ` Vincent MAILHOL
2022-06-21 15:03 ` Greg Kroah-Hartman
2022-06-21 15:54 ` Vincent MAILHOL
2022-06-21 15:11 ` Alan Stern
2022-06-21 15:55 ` Vincent MAILHOL
2022-06-21 16:14 ` Alan Stern
2022-06-21 16:40 ` Vincent MAILHOL
2022-06-21 17:00 ` Greg Kroah-Hartman
2022-06-21 17:14 ` Vincent MAILHOL
2022-06-21 17:46 ` Greg Kroah-Hartman
2022-06-22 9:22 ` David Laight
2022-06-22 9:41 ` Greg Kroah-Hartman
2022-06-22 10:03 ` David Laight
2022-06-22 11:11 ` Oliver Neukum
2022-06-22 10:34 ` Vincent MAILHOL
2022-06-22 12:23 ` Greg Kroah-Hartman [this message]
2022-06-22 15:59 ` Vincent MAILHOL
2022-06-22 18:11 ` Rhett Aultman
2022-06-26 8:21 ` Vincent MAILHOL
2022-06-27 19:24 ` Rhett Aultman
2022-06-28 1:09 ` Vincent MAILHOL
2022-07-04 13:02 ` Marc Kleine-Budde
2022-07-04 15:35 ` Rhett Aultman
2022-07-05 7:50 ` Marc Kleine-Budde
2022-06-23 17:30 ` Hongren Zenithal Zheng
2022-06-23 17:45 ` Shuah Khan
2022-06-24 14:43 ` Alan Stern
2022-06-24 16:01 ` Hongren Zenithal Zheng
2022-06-24 16:31 ` Shuah Khan
2022-06-24 18:07 ` Alan Stern
2022-06-27 22:54 ` Shuah Khan
2022-06-28 1:35 ` Alan Stern
2022-07-01 2:10 ` Shuah Khan
2022-08-01 17:42 ` Shuah Khan
2022-08-01 18:28 ` Vincent MAILHOL
2022-08-03 23:44 ` Shuah Khan
2022-06-10 21:33 ` [PATCH v3 2/2] can: gs_usb: fix DMA memory leak on close Rhett Aultman
2022-06-11 15:35 ` Marc Kleine-Budde
2022-06-11 16:03 ` Vincent MAILHOL
2022-06-12 21:28 ` David Laight
2022-06-12 21:33 ` Marc Kleine-Budde
2022-06-14 15:26 ` [PATCH v4 0/1] URB_FREE_COHERENT gs_usb memory leak fix Rhett Aultman
2022-06-14 15:26 ` [PATCH v4 1/1] can: gs_usb: fix DMA memory leak on close Rhett Aultman
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=YrMJ0vXR84cISTse@kroah.com \
--to=gregkh@linuxfoundation.org \
--cc=David.Laight@aculab.com \
--cc=linux-can@vger.kernel.org \
--cc=linux-usb@vger.kernel.org \
--cc=mailhol.vincent@wanadoo.fr \
--cc=mkl@pengutronix.de \
--cc=oneukum@suse.com \
--cc=rhett.aultman@samsara.com \
--cc=stern@rowland.harvard.edu \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).