Re: [PATCH 3/3] dma: mv_xor: use devm for dma pool and irq

[Rosen Penev]

On Wed, Jun 10, 2026 at 10:16 PM Vinod Koul <vkoul@xxxxxxxxxx> wrote:
>
> On 10-06-26, 09:42, Frank Li wrote:
> > On Tue, Jun 09, 2026 at 11:57:37PM -0700, Rosen Penev wrote:
> > > Replace dma_alloc_wc with dmam_alloc_attrs and request_irq
> > > with devm_request_irq. This eliminates the need for
> > > manual cleanup of the dma pool and irq in both the channel
> > > remove function and the channel add error labels, removing
> > > the err_free_irq and err_free_dma labels entirely.
> > >
> > > Assisted-by: opencode:big-pickle
> > > Signed-off-by: Rosen Penev <rosenp@xxxxxxxxx>
> > > ---
> >
> > I already said many times, tag should dmaengine, not dma, all functional
> > need (), please respect reviewer's time.
>
> And the worst is that some patches frpm Rosen have dmaengine, some dma.
> That make me wonder how much thought has been given to these changes and
> might be just an exercise to push AI generated code into wild and see
> what sticks
sort of. FWIW I use hardware that uses this driver:

 37:   89345094          0 GIC-0  54 Level     f1060800.xor
 38:  174555038          0 GIC-0  97 Level     f1060900.xor

Maybe all of this would finally fix

https://lore.kernel.org/all/CAKxU2N9yHa7ia_=07Csa7dDsZxcbPMmGSZKr+UxRSWH1VpG1fw@xxxxxxxxxxxxxx/

who knows...

Anyway, dma vs dmaengine is the model being stupid and me forgetting to fix it.

With other patches, I've gotten requests to try and fix pre existing
sashiko bug reports. I don't really want to but it is what it is.
>
> --
> ~Vinod