diff mbox series

[v7,RESEND,03/13] Revert "dmaengine: imx-sdma: fix context cache"

Message ID 1589218356-17475-4-git-send-email-yibin.gong@nxp.com (mailing list archive)
State New, archived
Headers show
Series add ecspi ERR009165 for i.mx6/7 soc family | expand

Commit Message

Robin Gong May 11, 2020, 5:32 p.m. UTC
This reverts commit d288bddd8374e0a043ac9dde64a1ae6a09411d74, since
'context_loaded' finally removed.

Signed-off-by: Robin Gong <yibin.gong@nxp.com>
---
 drivers/dma/imx-sdma.c | 1 -
 1 file changed, 1 deletion(-)

Comments

Sascha Hauer May 13, 2020, 6:05 a.m. UTC | #1
On Tue, May 12, 2020 at 01:32:26AM +0800, Robin Gong wrote:
> This reverts commit d288bddd8374e0a043ac9dde64a1ae6a09411d74, since
> 'context_loaded' finally removed.
> 
> Signed-off-by: Robin Gong <yibin.gong@nxp.com>
> ---
>  drivers/dma/imx-sdma.c | 1 -
>  1 file changed, 1 deletion(-)
> 
> diff --git a/drivers/dma/imx-sdma.c b/drivers/dma/imx-sdma.c
> index 4d4477d..3d4aac9 100644
> --- a/drivers/dma/imx-sdma.c
> +++ b/drivers/dma/imx-sdma.c
> @@ -1338,7 +1338,6 @@ static void sdma_free_chan_resources(struct dma_chan *chan)
>  
>  	sdmac->event_id0 = 0;
>  	sdmac->event_id1 = 0;
> -	sdmac->context_loaded = false;
>  
>  	sdma_set_channel_priority(sdmac, 0);

I think this can safely be folded into the next patch which makes it
more clear what is happening.

Sascha
Fuzzey, Martin May 13, 2020, 7:12 a.m. UTC | #2
On Wed, 13 May 2020 at 08:07, Sascha Hauer <s.hauer@pengutronix.de> wrote:
>
> On Tue, May 12, 2020 at 01:32:26AM +0800, Robin Gong wrote:
> > This reverts commit d288bddd8374e0a043ac9dde64a1ae6a09411d74, since
> > 'context_loaded' finally removed.
> >
> > Signed-off-by: Robin Gong <yibin.gong@nxp.com>
> > ---
>
> I think this can safely be folded into the next patch which makes it
> more clear what is happening.
>

Agreed,
not only that but having 2 separate patches also means that the bug
that was fixed by the commit being reverted could reappear during
bisection.

More generally I think reverts should be reserved for commits that
later turn out to be wrong or unneeded (ie should never really have
been applied).
If they were OK at the time but later become unnecessary due to other
code changes I think all the related modifications should be done in a
single normal non revert patch.

Martin
Robin Gong May 13, 2020, 8:06 a.m. UTC | #3
On 2020/05/13 Fuzzey, Martin <martin.fuzzey@flowbird.group> wrote:
> On Wed, 13 May 2020 at 08:07, Sascha Hauer <s.hauer@pengutronix.de>
> wrote:
> >
> > On Tue, May 12, 2020 at 01:32:26AM +0800, Robin Gong wrote:
> > > This reverts commit d288bddd8374e0a043ac9dde64a1ae6a09411d74, since
> > > 'context_loaded' finally removed.
> > >
> > > Signed-off-by: Robin Gong <yibin.gong@nxp.com>
> > > ---
> >
> > I think this can safely be folded into the next patch which makes it
> > more clear what is happening.
> >
> 
> Agreed,
> not only that but having 2 separate patches also means that the bug that was
> fixed by the commit being reverted could reappear during bisection.
> 
> More generally I think reverts should be reserved for commits that later turn
> out to be wrong or unneeded (ie should never really have been applied).
> If they were OK at the time but later become unnecessary due to other code
> changes I think all the related modifications should be done in a single normal
> non revert patch.
Okay, will remove it in v8.
diff mbox series

Patch

diff --git a/drivers/dma/imx-sdma.c b/drivers/dma/imx-sdma.c
index 4d4477d..3d4aac9 100644
--- a/drivers/dma/imx-sdma.c
+++ b/drivers/dma/imx-sdma.c
@@ -1338,7 +1338,6 @@  static void sdma_free_chan_resources(struct dma_chan *chan)
 
 	sdmac->event_id0 = 0;
 	sdmac->event_id1 = 0;
-	sdmac->context_loaded = false;
 
 	sdma_set_channel_priority(sdmac, 0);