On Thu, Sep 14, 2023 at 11:08 AM Yann Gautier <yann.gautier@xxxxxxxxxxx> wrote:
> On 9/4/23 14:21, Ulf Hansson wrote:
> > On Fri, 1 Sept 2023 at 16:10, Linus Walleij <linus.walleij@xxxxxxxxxx> wrote:
> >>
> >> Hi Yann/Christophe,
> >>
> >> thanks for your patch!
> >>
> >> On Fri, Sep 1, 2023 at 2:08 PM Yann Gautier <yann.gautier@xxxxxxxxxxx> wrote:
> >>
> >>> From: Christophe Kerello <christophe.kerello@xxxxxxxxxxx>
> >>>
> >>> Add the support of SDIO in-band interrupt mode for STM32 variant.
> >>> It allows the SD I/O card to interrupt the host on SDMMC_D1 data line.
> >>>
> >>> Signed-off-by: Christophe Kerello <christophe.kerello@xxxxxxxxxxx>
> >>> Signed-off-by: Yann Gautier <yann.gautier@xxxxxxxxxxx>
> >> (...)
> >>> +++ b/drivers/mmc/host/mmci.h
> >>> @@ -332,6 +332,7 @@ enum mmci_busy_state {
> >>> * @opendrain: bitmask identifying the OPENDRAIN bit inside MMCIPOWER register
> >>> * @dma_lli: true if variant has dma link list feature.
> >>> * @stm32_idmabsize_mask: stm32 sdmmc idma buffer size.
> >>> + * @use_sdio_irq: allow SD I/O card to interrupt the host
> >>
> >> The documentation tag should be one line up (compare to the members...)
> >>
> >>> @@ -376,6 +377,7 @@ struct variant_data {
> >>> u32 start_err;
> >>> u32 opendrain;
> >>> u8 dma_lli:1;
> >>> + u8 use_sdio_irq:1;
> >>
> >> 1. bool use_sdio_irq;
> >>
> Hi,
>
> Should it really be changed to a bool?
> Other boolean likes in the structure are u8:1.
Yes, two wrongs does not make one right.
Using u8:1 is a way of trying to outsmart the compiler
which is generally a bad idea.
> > That said, the out-of-band IRQs is what works best for the ux500 variants.
>
> What I understand here is that in-band interrupts are not properly
> working on ux500, and then the feature shouldn't be enabled for this
> platform.
> Am I correct?
I think we can flag the feature as available and implement the handling
but add a comment that this is unstable and that Ux500 users should
prefer to use out-of-band IRQs.
Yours,
Linus Walleij
