| Message ID | ef6f884ef144390f152c34d2f549b1f50303b7b1.1658262447.git.christophe.jaillet@wanadoo.fr (mailing list archive) |
|---|---|
| State | Superseded |
| Delegated to: | Herbert Xu |
| Headers | show |
| Series | crypto: ccree - Remove a useless dma_supported() call | expand |
On Tue, Jul 19, 2022 at 10:27:50PM +0200, Christophe JAILLET wrote: > There is no point in calling dma_supported() before calling > dma_set_coherent_mask(). This function already calls dma_supported() and > returns an error (-EIO) if it fails. > > So remove the superfluous dma_supported() call. > > While at it, fix the name of the function reported in a dev_err(). > > Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> > --- > I guess that the whole while loop could be removed, but I don't remind the > thread with the corresponding explanation, so leave it as-is :( The loop should go away. Setting a larger DMA mask will never fail when setting a smaller one will succeed. Also after this patch dma_supported can be marked static (Yay!)
Hi Christophe, Thank you for the patch! On Tue, Jul 19, 2022 at 11:27 PM Christophe JAILLET <christophe.jaillet@wanadoo.fr> wrote: > > There is no point in calling dma_supported() before calling > dma_set_coherent_mask(). This function already calls dma_supported() and > returns an error (-EIO) if it fails. > > So remove the superfluous dma_supported() call. > > While at it, fix the name of the function reported in a dev_err(). > > Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> Acked-by: Gilad Ben-Yossef <gilad@benyossef.com> > --- > I guess that the whole while loop could be removed, but I don't remind the > thread with the corresponding explanation, so leave it as-is :( I would be happy to ack a patch that does this if you care to write it... Gilad > --- > drivers/crypto/ccree/cc_driver.c | 12 ++++++------ > 1 file changed, 6 insertions(+), 6 deletions(-) > > diff --git a/drivers/crypto/ccree/cc_driver.c b/drivers/crypto/ccree/cc_driver.c > index 7d1bee86d581..99f8bda718fe 100644 > --- a/drivers/crypto/ccree/cc_driver.c > +++ b/drivers/crypto/ccree/cc_driver.c > @@ -373,16 +373,16 @@ static int init_cc_resources(struct platform_device *plat_dev) > > dma_mask = DMA_BIT_MASK(DMA_BIT_MASK_LEN); > while (dma_mask > 0x7fffffffUL) { > - if (dma_supported(dev, dma_mask)) { > - rc = dma_set_coherent_mask(dev, dma_mask); > - if (!rc) > - break; > - } > + rc = dma_set_coherent_mask(dev, dma_mask); > + if (!rc) > + break; > + > dma_mask >>= 1; > } > > if (rc) { > - dev_err(dev, "Failed in dma_set_mask, mask=%llx\n", dma_mask); > + dev_err(dev, "Failed in dma_set_coherent_mask, mask=%llx\n", > + dma_mask); > return rc; > } > > -- > 2.34.1 >
Le 20/07/2022 à 11:24, Gilad Ben-Yossef a écrit : > Hi Christophe, > > Thank you for the patch! > > > On Tue, Jul 19, 2022 at 11:27 PM Christophe JAILLET > <christophe.jaillet@wanadoo.fr> wrote: >> >> There is no point in calling dma_supported() before calling >> dma_set_coherent_mask(). This function already calls dma_supported() and >> returns an error (-EIO) if it fails. >> >> So remove the superfluous dma_supported() call. >> >> While at it, fix the name of the function reported in a dev_err(). >> >> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> > > Acked-by: Gilad Ben-Yossef <gilad@benyossef.com> > >> --- >> I guess that the whole while loop could be removed, but I don't remind the >> thread with the corresponding explanation, so leave it as-is :( > > I would be happy to ack a patch that does this if you care to write it... I will send a v2. Don't bother with this patch. CJ > > Gilad > >> --- >> drivers/crypto/ccree/cc_driver.c | 12 ++++++------ >> 1 file changed, 6 insertions(+), 6 deletions(-) >> >> diff --git a/drivers/crypto/ccree/cc_driver.c b/drivers/crypto/ccree/cc_driver.c >> index 7d1bee86d581..99f8bda718fe 100644 >> --- a/drivers/crypto/ccree/cc_driver.c >> +++ b/drivers/crypto/ccree/cc_driver.c >> @@ -373,16 +373,16 @@ static int init_cc_resources(struct platform_device *plat_dev) >> >> dma_mask = DMA_BIT_MASK(DMA_BIT_MASK_LEN); >> while (dma_mask > 0x7fffffffUL) { >> - if (dma_supported(dev, dma_mask)) { >> - rc = dma_set_coherent_mask(dev, dma_mask); >> - if (!rc) >> - break; >> - } >> + rc = dma_set_coherent_mask(dev, dma_mask); >> + if (!rc) >> + break; >> + >> dma_mask >>= 1; >> } >> >> if (rc) { >> - dev_err(dev, "Failed in dma_set_mask, mask=%llx\n", dma_mask); >> + dev_err(dev, "Failed in dma_set_coherent_mask, mask=%llx\n", >> + dma_mask); >> return rc; >> } >> >> -- >> 2.34.1 >> > >
diff --git a/drivers/crypto/ccree/cc_driver.c b/drivers/crypto/ccree/cc_driver.c index 7d1bee86d581..99f8bda718fe 100644 --- a/drivers/crypto/ccree/cc_driver.c +++ b/drivers/crypto/ccree/cc_driver.c @@ -373,16 +373,16 @@ static int init_cc_resources(struct platform_device *plat_dev) dma_mask = DMA_BIT_MASK(DMA_BIT_MASK_LEN); while (dma_mask > 0x7fffffffUL) { - if (dma_supported(dev, dma_mask)) { - rc = dma_set_coherent_mask(dev, dma_mask); - if (!rc) - break; - } + rc = dma_set_coherent_mask(dev, dma_mask); + if (!rc) + break; + dma_mask >>= 1; } if (rc) { - dev_err(dev, "Failed in dma_set_mask, mask=%llx\n", dma_mask); + dev_err(dev, "Failed in dma_set_coherent_mask, mask=%llx\n", + dma_mask); return rc; }
There is no point in calling dma_supported() before calling dma_set_coherent_mask(). This function already calls dma_supported() and returns an error (-EIO) if it fails. So remove the superfluous dma_supported() call. While at it, fix the name of the function reported in a dev_err(). Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> --- I guess that the whole while loop could be removed, but I don't remind the thread with the corresponding explanation, so leave it as-is :( --- drivers/crypto/ccree/cc_driver.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)