| Message ID | 20230814115523.15279-1-tiwai@suse.de |
|---|---|
| Headers | show |
| Series | ALSA: Generic PCM copy ops using iov_iter | expand |
On Mon, 14 Aug 2023 14:15:18 +0200, Ismael Luceno wrote: > > On 14/Aug/2023 13:55, Takashi Iwai wrote: > > This patch converts the solo6x10 driver code to use the new unified > > PCM copy callback. It's a straightforward conversion from *_user() to > > *_iter() variants. As copy_to_iter() updates the internal offest at > > each write, we can drop the dst counter update in the loop, too. > > > > Note that copy_from/to_iter() returns the copied bytes, hence the > > error condition is inverted from copy_from/to_user(). > > > > Cc: Bluecherry Maintainers <maintainers@bluecherrydvr.com> > > Cc: Anton Sviridenko <anton@corp.bluecherry.net> > > Cc: Andrey Utkin <andrey_utkin@fastmail.com> > > Cc: Ismael Luceno <ismael@iodev.co.uk> > > Cc: Mauro Carvalho Chehab <mchehab@kernel.org> > > Cc: linux-media@vger.kernel.org > > Signed-off-by: Takashi Iwai <tiwai@suse.de> > > --- > > drivers/media/pci/solo6x10/solo6x10-g723.c | 38 +++------------------- > > 1 file changed, 5 insertions(+), 33 deletions(-) > > > > diff --git a/drivers/media/pci/solo6x10/solo6x10-g723.c b/drivers/media/pci/solo6x10/solo6x10-g723.c > > index 6cebad665565..aceacb822cab 100644 > > --- a/drivers/media/pci/solo6x10/solo6x10-g723.c > > +++ b/drivers/media/pci/solo6x10/solo6x10-g723.c > > @@ -204,9 +204,9 @@ static snd_pcm_uframes_t snd_solo_pcm_pointer(struct snd_pcm_substream *ss) > > return idx * G723_FRAMES_PER_PAGE; > > } > > > > -static int snd_solo_pcm_copy_user(struct snd_pcm_substream *ss, int channel, > > - unsigned long pos, void __user *dst, > > - unsigned long count) > > +static int snd_solo_pcm_copy(struct snd_pcm_substream *ss, int channel, > > + unsigned long pos, struct iov_iter *dst, > > + unsigned long count) > > { > > struct solo_snd_pcm *solo_pcm = snd_pcm_substream_chip(ss); > > struct solo_dev *solo_dev = solo_pcm->solo_dev; > > @@ -223,35 +223,8 @@ static int snd_solo_pcm_copy_user(struct snd_pcm_substream *ss, int channel, > > if (err) > > return err; > > > > - if (copy_to_user(dst, solo_pcm->g723_buf, G723_PERIOD_BYTES)) > > + if (!copy_to_iter(solo_pcm->g723_buf, G723_PERIOD_BYTES, dst)) > > return -EFAULT; > > - dst += G723_PERIOD_BYTES; > > - } > > - > > - return 0; > > -} > > - > > -static int snd_solo_pcm_copy_kernel(struct snd_pcm_substream *ss, int channel, > > - unsigned long pos, void *dst, > > - unsigned long count) > > -{ > > - struct solo_snd_pcm *solo_pcm = snd_pcm_substream_chip(ss); > > - struct solo_dev *solo_dev = solo_pcm->solo_dev; > > - int err, i; > > - > > - for (i = 0; i < (count / G723_FRAMES_PER_PAGE); i++) { > > - int page = (pos / G723_FRAMES_PER_PAGE) + i; > > - > > - err = solo_p2m_dma_t(solo_dev, 0, solo_pcm->g723_dma, > > - SOLO_G723_EXT_ADDR(solo_dev) + > > - (page * G723_PERIOD_BLOCK) + > > - (ss->number * G723_PERIOD_BYTES), > > - G723_PERIOD_BYTES, 0, 0); > > - if (err) > > - return err; > > - > > - memcpy(dst, solo_pcm->g723_buf, G723_PERIOD_BYTES); > > - dst += G723_PERIOD_BYTES; > > } > > > > return 0; > > @@ -263,8 +236,7 @@ static const struct snd_pcm_ops snd_solo_pcm_ops = { > > .prepare = snd_solo_pcm_prepare, > > .trigger = snd_solo_pcm_trigger, > > .pointer = snd_solo_pcm_pointer, > > - .copy_user = snd_solo_pcm_copy_user, > > - .copy_kernel = snd_solo_pcm_copy_kernel, > > + .copy = snd_solo_pcm_copy, > > }; > > > > static int snd_solo_capture_volume_info(struct snd_kcontrol *kcontrol, > > -- > > 2.35.3 > > > > Signed-off-by: Ismael Luceno <ismael@iodev.co.uk> You meant Reviewed-by or Acked-by? Signed-off-by is a tag used when you carry a patch. thanks, Takashi
On Mon, Aug 14, 2023 at 01:55:18PM +0200, Takashi Iwai wrote: > This patch converts the qcom lpass driver code to use the new unified > PCM copy callback. It's a straightforward conversion from *_user() to > *_iter() variants. > > Note that copy_from/to_iter() returns the copied bytes, hence the > error condition is inverted from copy_from/to_user(). Reviewed-by: Mark Brown <broonie@kernel.org>
On Mon, Aug 14, 2023 at 01:55:20PM +0200, Takashi Iwai wrote: > Along with the conversion to PCM copy ops, use the iov_iter for the > pointer to be passed to the dmaengine process callback, too. It > avoids the direct reference of iter_iov_addr(), and it can potentially > help for the drivers to access memory properly (although both atmel > and stm drivers don't use the given buffer address at all for now). Reviewed-by: Mark Brown <broonie@kernel.org>
On Mon, Aug 14, 2023 at 01:55:01PM +0200, Takashi Iwai wrote: > Add two more helpers for copying memory between iov_iter and iomem, > which will be used by the new PCM copy ops in a few drivers. > The existing helpers became wrappers of those now. > > Note that copy_from/to_iter() returns the copied bytes, hence the > error condition is inverted from copy_from/to_user(). ... > - return copy_to_user(dst, (const void __force*)src, count) ? -EFAULT : 0; > + return !copy_to_iter((const void __force *)src, count, dst) ? -EFAULT : 0; Inverted means, I believe, this return copy_to_iter((const void __force *)src, count, dst) == count ? 0 : -EFAULT; as far as I understand the idea behind the copy_to_user() return value. Ditto for other cases.
On Mon, Aug 14, 2023 at 01:55:02PM +0200, Takashi Iwai wrote: > This patch converts the dummy driver code to use the new unified PCM > copy callback. As dummy driver doesn't do anything in the callback, > it's just a simple replacement. Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
On Tue, 15 Aug 2023 15:51:30 +0200, Andy Shevchenko wrote: > > On Mon, Aug 14, 2023 at 01:55:01PM +0200, Takashi Iwai wrote: > > Add two more helpers for copying memory between iov_iter and iomem, > > which will be used by the new PCM copy ops in a few drivers. > > The existing helpers became wrappers of those now. > > > > Note that copy_from/to_iter() returns the copied bytes, hence the > > error condition is inverted from copy_from/to_user(). > > ... > > > - return copy_to_user(dst, (const void __force*)src, count) ? -EFAULT : 0; > > + return !copy_to_iter((const void __force *)src, count, dst) ? -EFAULT : 0; > > Inverted means, I believe, this > > return copy_to_iter((const void __force *)src, count, dst) == count ? 0 : -EFAULT; > > as far as I understand the idea behind the copy_to_user() return value. > > Ditto for other cases. OK, this looks more correct. I'll fix in v2. thanks, Takashi
On Mon, Aug 14, 2023 at 01:55:06PM +0200, Takashi Iwai wrote: > This patch converts the korg1212 driver code to use the new unified > PCM copy callback. The open-coded conditional memory copies are > replaced with simpler copy_from/to_iter() calls. > > Note that copy_from/to_iter() returns the copied bytes, hence the > error condition is inverted from copy_from/to_user(). ... > + if (!copy_to_iter(src, size, dst)) != size ? > return -EFAULT; ... > + if (!copy_from_iter(dst, size, src)) > return -EFAULT; Ditto.