alsa-devel.alsa-project.org archive mirror
 help / color / mirror / Atom feed
* USB transfer_buffer allocations on 64bit systems
@ 2010-04-07  9:06 Daniel Mack
       [not found] ` <p2g581ef6d61004070220z1153d40ez955b356e01220848@mail.gmail.com>
  2010-04-07 14:59 ` USB transfer_buffer allocations on 64bit systems Alan Stern
  0 siblings, 2 replies; 104+ messages in thread
From: Daniel Mack @ 2010-04-07  9:06 UTC (permalink / raw)
  To: linux-kernel
  Cc: alsa-devel, linux-usb, Greg KH, Alan Stern, Pedro Ribeiro, akpm

Hi,

I was pointed to https://bugzilla.kernel.org/show_bug.cgi?id=15580 by
Pedro Ribeiro, and unfortunately I'm pretty late in the game. I wasn't
aware of that thread until yesterday.

While the report is quite confusing, the reason seams pretty clear to me
as I've been thru quite some time-demanding debugging of a very similar
issue on Mac OS X. But I'm not totally sure whether we really hit the
same issue here, so I'd like to have your opinions first.

The problem is appearantly the way the transfer buffer is allocated in
the drivers. In the snd-usb-caiaq driver, I used kzalloc() to get memory
which works fine on 32bit systems. On x86_64, however, it seems that
kzalloc() hands out memory beyond the 32bit addressable boundary, which
the DMA controller of the 32bit PCI-connected EHCI controller is unable
to write to or read from. Am I correct on this conclusion?

Depending on the condition of the memory management, things might work
or not, and especially right after a reboot, there's a better chance to
get lower memory.

The fix is to use usb_buffer_alloc() for that purpose which ensures
memory that is suitable for DMA. And on x86_64, this also means that the
upper 32 bits of the address returned are all 0's.

If what I've stated is true, there are quite some more drivers affected
by this issue. I collected a list of places where similar fixes are
needed, and I can send patches if I get a thumbs-up.

Pedro is currently testing a patch I sent out yesterday.

Thanks,
Daniel

^ permalink raw reply	[flat|nested] 104+ messages in thread
[parent not found: <g2h74fd948d1004141820ladc1941eo732d059dd678df0a@mail.gmail.com>]
* Re: USB transfer_buffer allocations on 64bit systems
  2010-05-07 11:47           ` Oliver Neukum
@ 2010-05-07 11:58 Daniel Mack
       [not found] ` <20100507115810.GN30801-ahpEBR4enfnCULTFXS99ULNAH6kLmebB@public.gmane.org>
  -1 siblings, 1 reply; 104+ messages in thread
From: Daniel Mack @ 2010-05-07 11:58 UTC (permalink / raw)
  To: Oliver Neukum
  Cc: alsa-devel, USB list, Konrad Rzeszutek Wilk, Takashi Iwai,
	Greg KH, Clemens Ladisch, Kernel development list, Chris Wright,
	iommu, Andi Kleen, Alan Stern, Pedro Ribeiro, Andrew Morton,
	David Woodhouse

On Fri, May 07, 2010 at 01:47:56PM +0200, Oliver Neukum wrote:
> Am Freitag, 7. Mai 2010 13:42:34 schrieb Oliver Neukum:
> > Am Freitag, 7. Mai 2010 11:47:37 schrieb Clemens Ladisch:
> > > > In the particular case of audio drivers, though, the contents of the
> > > > buffers are likely to change after the submission. What we do here
> > > > is that we map the audio stream buffers which are used by ALSA to
> > > > the output URBs, so they're filled asychronously. Once the buffer is
> > > > actually sent out on the bus, it is believed to contain proper audio
> > > > date. If it doesn't, that's due to too tight audio timing or other
> > > > problems. This breaks once buffers are magically bounced in the
> > > > background.
> > > 
> > > At least the audio class and ua101 drivers don't do this and fill the
> > > buffers before they are submitted.
> > > 
> > > > So - long story short: these audio buffers need to be DMA coherent.
> > > 
> > > Does the USB API actually guarantee that all controllers use DMA, i.e.,
> > > that the buffers can be filled after submission?
> > 
> > No, you must not touch buffers after submission. This does not even
> > work if we use DMA, because on some architectures this violates
> > guarantees to the dma primitives. It cannot be done.

We do such tricks on other OS where IRQ latency as high as some tens
of milliseconds. So we queue EHCI transfers well in advance and update
access their contents (for both input and output) from audio context.
I considered implementing this idea to the Linux USB audio driver as
well.

> Or to be precise it could be done with coherent memory, but you'd risk
> transfering partially updated buffers, as you cannot know when DMA will
> be done and we don't guarantee that DMA will be done right as we transfer.

If the memory is coherent, it should be possible. And if buffers are
only partially updated, the audio buffer scheduling is too tight. IOW,
the configured buffer size of audio clients in userspace is too small.

This is OT now I believe, I will start another thread for this topic
once I have something to show.

Thanks,
Daniel

^ permalink raw reply	[flat|nested] 104+ messages in thread
* Re: USB transfer_buffer allocations on 64bit systems
  2010-05-10  2:50         ` FUJITA Tomonori
@ 2010-05-10  9:21 David Woodhouse
       [not found] ` <1273483265.372.3383.camel-uXGAPMMVk8bAQYKIod7YupZV94DADvEd@public.gmane.org>
  -1 siblings, 1 reply; 104+ messages in thread
From: David Woodhouse @ 2010-05-10  9:21 UTC (permalink / raw)
  To: FUJITA Tomonori
  Cc: alsa-devel, linux-usb, konrad.wilk, tiwai, gregkh, clemens,
	linux-kernel, chrisw, iommu, andi, stern, pedrib, akpm

On Mon, 2010-05-10 at 11:50 +0900, FUJITA Tomonori wrote:
> On Fri, 7 May 2010 10:51:10 -0400 (EDT)
> Alan Stern <stern@rowland.harvard.edu> wrote:
> 
> > On Fri, 7 May 2010, Daniel Mack wrote:
> > 
> > > > At least the audio class and ua101 drivers don't do this and fill the
> > > > buffers before they are submitted.
> > > 
> > > Gnaa, you're right. I _thought_ my code does it the way I described, but
> > > what I wrote is how I _wanted_ to do it, not how it's currently done. I
> > > have a plan to change this in the future.
> > > 
> > > So unfortunately, that doesn't explain it either. Sorry for the noise.
> > 
> > At one point we tried an experiment, printing out the buffer and DMA 
> > addresses.  I don't recall seeing anything obviously wrong, but if an 
> > IOMMU was in use then that might not mean anything.  Is it possible 
> > that the IOMMU mappings sometimes get messed up for addresses above 4 
> > GB?
> 
> You mean that an IOMMU could allocate an address above 4GB wrongly? If
> so, IIRC, all the IOMMU implementations use dev->dma_mask and
> dev->coherent_dma_mask properly. And the DMA address space of the
> majority of IOMMUs are limited less than 4GB.

The Intel IOMMU code will use dev->dma_mask and dev->coherent_dma_mask
properly. It is not limited to 4GiB, but it will tend to give virtual
DMA addresses below 4GiB even when a device is capable of more; it'll
only give out higher addresses when the address space below 4GiB is
exhausted.

-- 
dwmw2

^ permalink raw reply	[flat|nested] 104+ messages in thread

end of thread, other threads:[~2010-05-11 15:34 UTC | newest]

Thread overview: 104+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2010-04-07  9:06 USB transfer_buffer allocations on 64bit systems Daniel Mack
     [not found] ` <p2g581ef6d61004070220z1153d40ez955b356e01220848@mail.gmail.com>
2010-04-07  9:26   ` USB HID gadget driver (was: Re: USB transfer_buffer allocations on 64bit systems) Daniel Mack
2010-04-07 14:59 ` USB transfer_buffer allocations on 64bit systems Alan Stern
2010-04-07 15:11   ` Daniel Mack
2010-04-07 15:31     ` Greg KH
2010-04-07 15:35       ` Daniel Mack
2010-04-07 15:51         ` Greg KH
     [not found]           ` <20100407155122.GA13974-U8xfFu+wG4EAvxtiuMwx3w@public.gmane.org>
2010-04-07 16:04             ` Alan Stern
2010-04-08  6:09         ` Oliver Neukum
     [not found]           ` <201004080809.11756.oliver-GvhC2dPhHPQdnm+yROfE0A@public.gmane.org>
2010-04-08 11:07             ` Daniel Mack
2010-04-07 15:55       ` Alan Stern
2010-04-07 16:16         ` Daniel Mack
2010-04-07 16:47           ` Alan Stern
2010-04-07 17:55           ` Takashi Iwai
2010-04-07 17:59             ` Daniel Mack
2010-04-07 18:06               ` Takashi Iwai
2010-04-07 19:13             ` Alan Stern
2010-04-08  0:33               ` Greg KH
2010-04-09  0:01                 ` Robert Hancock
     [not found]                   ` <4BBE6E57.6020600-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
2010-04-09 16:50                     ` Sarah Sharp
2010-04-09 23:38                       ` Robert Hancock
2010-04-10  8:34                         ` Daniel Mack
2010-04-10 17:02                           ` [alsa-devel] " Robert Hancock
2010-04-12 18:56                             ` Sarah Sharp
2010-04-12 20:39                               ` Robert Hancock
2010-04-12 20:58                                 ` Sarah Sharp
     [not found]               ` <Pine.LNX.4.44L0.1004071452560.5760-100000-IYeN2dnnYyZXsRXLowluHWD2FQJk+8+b@public.gmane.org>
2010-04-07 23:59                 ` Robert Hancock
2010-04-12 11:17                 ` [PATCH] USB: rename usb_buffer_alloc() and usb_buffer_free() Daniel Mack
     [not found]                   ` <1271071045-3112-1-git-send-email-daniel-rDUAYElUppE@public.gmane.org>
2010-04-13 18:16                     ` Daniel Mack
     [not found]                       ` <20100413181631.GT30801-ahpEBR4enfnCULTFXS99ULNAH6kLmebB@public.gmane.org>
2010-04-13 19:27                         ` Alan Stern
2010-04-13 20:26                           ` Greg KH
2010-04-13 21:47                             ` Daniel Mack
2010-04-07 17:52         ` USB transfer_buffer allocations on 64bit systems Takashi Iwai
2010-04-07 15:46     ` Alan Stern
2010-04-08  6:12       ` Oliver Neukum
     [not found]         ` <201004080812.04419.oliver-GvhC2dPhHPQdnm+yROfE0A@public.gmane.org>
2010-04-08 16:59           ` Alan Stern
2010-04-08 21:24             ` Oliver Neukum
2010-04-08 22:20               ` Alan Stern
2010-04-09  6:04                 ` Oliver Neukum
     [not found]                   ` <201004090804.36213.oliver-GvhC2dPhHPQdnm+yROfE0A@public.gmane.org>
2010-04-09 14:41                     ` Alan Stern
     [not found]                       ` <Pine.LNX.4.44L0.1004091033150.1852-100000-IYeN2dnnYyZXsRXLowluHWD2FQJk+8+b@public.gmane.org>
2010-04-09 14:50                         ` Oliver Neukum
     [not found]                           ` <201004091650.31488.oliver-GvhC2dPhHPQdnm+yROfE0A@public.gmane.org>
2010-04-09 15:15                             ` Alan Stern
     [not found]                               ` <Pine.LNX.4.44L0.1004091114500.1852-100000-IYeN2dnnYyZXsRXLowluHWD2FQJk+8+b@public.gmane.org>
2010-04-09 20:51                                 ` Oliver Neukum
2010-04-09 21:21                                   ` Alan Stern
     [not found]   ` <Pine.LNX.4.44L0.1004071036060.1779-100000-IYeN2dnnYyZXsRXLowluHWD2FQJk+8+b@public.gmane.org>
2010-04-07 16:54     ` Oliver Neukum
     [not found]       ` <201004071854.55530.oliver-GvhC2dPhHPQdnm+yROfE0A@public.gmane.org>
2010-04-07 17:00         ` Daniel Mack
2010-04-07 23:55     ` Robert Hancock
     [not found]       ` <4BBD1B6F.3000205-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
2010-04-08  2:10         ` Alan Stern
2010-04-08  7:30           ` Daniel Mack
     [not found]             ` <20100408073041.GO30801-ahpEBR4enfnCULTFXS99ULNAH6kLmebB@public.gmane.org>
2010-04-08 16:57               ` Alan Stern
2010-04-08 17:17                 ` Pedro Ribeiro
     [not found]                 ` <Pine.LNX.4.44L0.1004081245330.1720-100000-IYeN2dnnYyZXsRXLowluHWD2FQJk+8+b@public.gmane.org>
2010-04-08 23:13                   ` Pedro Ribeiro
2010-04-09 16:01                     ` Alan Stern
2010-04-09 18:09                       ` Daniel Mack
     [not found]                         ` <20100409180942.GK30801-ahpEBR4enfnCULTFXS99ULNAH6kLmebB@public.gmane.org>
2010-04-09 18:19                           ` Pedro Ribeiro
     [not found]                             ` <w2r74fd948d1004091119j9f33d8a6kc1824d9243abf38b-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2010-04-09 19:34                               ` Alan Stern
2010-04-09 20:14                                 ` Daniel Mack
     [not found]                                 ` <Pine.LNX.4.44L0.1004091529240.1852-100000-IYeN2dnnYyZXsRXLowluHWD2FQJk+8+b@public.gmane.org>
2010-04-09 20:25                                   ` [LKML] " Konrad Rzeszutek Wilk
     [not found]                                     ` <20100409202533.GA8983-6K5HmflnPlqSPmnEAIUT9EEOCMrvLtNR@public.gmane.org>
2010-04-09 21:23                                       ` Alan Stern
2010-04-09 22:11                                         ` Robert Hancock
2010-04-12 10:48                                           ` Daniel Mack
2010-04-12 12:06                                             ` Pedro Ribeiro
2010-04-10 12:49                             ` Daniel Mack
     [not found]                               ` <20100410124912.GP30801-ahpEBR4enfnCULTFXS99ULNAH6kLmebB@public.gmane.org>
2010-04-10 13:21                                 ` Pedro Ribeiro
2010-04-12  8:59   ` Andi Kleen
2010-04-12 11:14     ` Daniel Mack
     [not found]       ` <20100412111439.GU30801-ahpEBR4enfnCULTFXS99ULNAH6kLmebB@public.gmane.org>
2010-04-12 11:53         ` Andi Kleen
2010-04-12 12:11           ` Pedro Ribeiro
     [not found]             ` <q2z74fd948d1004120511hebf19eaauc41ee9ed25dea19e-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2010-04-12 12:12               ` Andi Kleen
2010-04-12 12:32                 ` Daniel Mack
2010-04-12 12:47                   ` Andi Kleen
2010-04-12 12:54                     ` Daniel Mack
2010-04-12 15:43                       ` Andi Kleen
     [not found]                         ` <20100412154323.GP18855-qrUzlfsMFqo/4alezvVtWx2eb7JE58TQ@public.gmane.org>
2010-04-12 16:17                           ` Alan Stern
2010-04-12 16:29                             ` Andi Kleen
2010-04-12 16:57                               ` Alan Stern
2010-04-12 17:15                                 ` Daniel Mack
     [not found]                                   ` <20100412171507.GB30801-ahpEBR4enfnCULTFXS99ULNAH6kLmebB@public.gmane.org>
2010-04-12 17:22                                     ` Andi Kleen
2010-04-12 17:56                                       ` Daniel Mack
2010-04-12 17:52                                     ` [LKML] " Konrad Rzeszutek Wilk
2010-04-13 18:22                                 ` Daniel Mack
     [not found]                                   ` <20100413182233.GR30807-ahpEBR4enfnCULTFXS99ULNAH6kLmebB@public.gmane.org>
2010-04-13 23:46                                     ` Pedro Ribeiro
2010-04-14 10:09                                       ` Daniel Mack
2010-04-14 10:47                                         ` Pedro Ribeiro
2010-04-14 11:02                                           ` Pedro Ribeiro
2010-04-14 13:18                                           ` [LKML] " Konrad Rzeszutek Wilk
     [not found]                                           ` <t2w74fd948d1004140347k3447bffapb73856eacddfde55-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2010-04-14 14:08                                             ` Alan Stern
2010-04-14 16:36                                               ` Daniel Mack
     [not found]                                                 ` <20100414163637.GV30807-ahpEBR4enfnCULTFXS99ULNAH6kLmebB@public.gmane.org>
2010-04-14 17:21                                                   ` Pedro Ribeiro
2010-04-15  7:35                                                     ` Daniel Mack
     [not found] <g2h74fd948d1004141820ladc1941eo732d059dd678df0a@mail.gmail.com>
     [not found] ` <Pine.LNX.4.44L0.1004151114490.1562-100000@iolanthe.rowland.org>
2010-05-07  7:48 ` Daniel Mack
2010-05-07  9:47   ` Clemens Ladisch
2010-05-07 10:24     ` Daniel Mack
2010-05-07 14:51       ` [alsa-devel] " Alan Stern
2010-05-10  2:50         ` FUJITA Tomonori
     [not found]       ` <20100507102408.GM30801-ahpEBR4enfnCULTFXS99ULNAH6kLmebB@public.gmane.org>
2010-05-10 14:31         ` Konrad Rzeszutek Wilk
     [not found]     ` <4BE3E1B9.5020602-P6GI/4k7KOmELgA04lAiVw@public.gmane.org>
2010-05-07 11:42       ` Oliver Neukum
     [not found]         ` <201005071342.34923.oneukum-l3A5Bk7waGM@public.gmane.org>
2010-05-07 11:47           ` Oliver Neukum
  -- strict thread matches above, loose matches on Subject: below --
2010-05-07 11:58 Daniel Mack
     [not found] ` <20100507115810.GN30801-ahpEBR4enfnCULTFXS99ULNAH6kLmebB@public.gmane.org>
2010-05-07 14:45   ` [alsa-devel] " Alan Stern
2010-05-10  9:21 David Woodhouse
     [not found] ` <1273483265.372.3383.camel-uXGAPMMVk8bAQYKIod7YupZV94DADvEd@public.gmane.org>
2010-05-10 14:58   ` [alsa-devel] " Alan Stern
     [not found]     ` <Pine.LNX.4.44L0.1005101049100.1626-100000-IYeN2dnnYyZXsRXLowluHWD2FQJk+8+b@public.gmane.org>
2010-05-11  1:06       ` FUJITA Tomonori
     [not found]         ` <20100511100637D.fujita.tomonori-Zyj7fXuS5i5L9jVzuh4AOg@public.gmane.org>
2010-05-11 14:00           ` Alan Stern
2010-05-11 14:22             ` FUJITA Tomonori
2010-05-11 14:24             ` Konrad Rzeszutek Wilk
2010-05-11 14:38               ` FUJITA Tomonori
2010-05-11 15:04                 ` Alan Stern
2010-05-11 15:34                   ` FUJITA Tomonori

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).