2024-02-20 23:01:33

by Linus Walleij

Subject: [PATCH] mmc: sh_mmcif: sg_miter does not need to be atomic

All the sglist iterations happen in the *threaded* interrupt handler
and that context is not atomic, so don't request an atomic
sglist miter. Using an atomic miter results in "BUG: scheduling while
atomic" splats.

Reported-by: Geert Uytterhoeven <[email protected]>
Fixes: 27b57277d9ba ("mmc: sh_mmcif: Use sg_miter for PIO")
Signed-off-by: Linus Walleij <[email protected]>
---
Hi Geert, it'd be great if you could test this!
---
drivers/mmc/host/sh_mmcif.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/mmc/host/sh_mmcif.c b/drivers/mmc/host/sh_mmcif.c
index 1ef6e153e5a3..669555b5e8fa 100644
--- a/drivers/mmc/host/sh_mmcif.c
+++ b/drivers/mmc/host/sh_mmcif.c
@@ -607,7 +607,7 @@ static void sh_mmcif_single_read(struct sh_mmcif_host *host,
BLOCK_SIZE_MASK) + 3;

sg_miter_start(&host->sg_miter, data->sg, data->sg_len,
- SG_MITER_ATOMIC | SG_MITER_TO_SG);
+ SG_MITER_TO_SG);

host->wait_for = MMCIF_WAIT_FOR_READ;

@@ -662,7 +662,7 @@ static void sh_mmcif_multi_read(struct sh_mmcif_host *host,
BLOCK_SIZE_MASK;

sg_miter_start(&host->sg_miter, data->sg, data->sg_len,
- SG_MITER_ATOMIC | SG_MITER_TO_SG);
+ SG_MITER_TO_SG);

host->wait_for = MMCIF_WAIT_FOR_MREAD;

@@ -710,7 +710,7 @@ static void sh_mmcif_single_write(struct sh_mmcif_host *host,
BLOCK_SIZE_MASK) + 3;

sg_miter_start(&host->sg_miter, data->sg, data->sg_len,
- SG_MITER_ATOMIC | SG_MITER_FROM_SG);
+ SG_MITER_FROM_SG);

host->wait_for = MMCIF_WAIT_FOR_WRITE;

@@ -765,7 +765,7 @@ static void sh_mmcif_multi_write(struct sh_mmcif_host *host,
BLOCK_SIZE_MASK;

sg_miter_start(&host->sg_miter, data->sg, data->sg_len,
- SG_MITER_ATOMIC | SG_MITER_FROM_SG);
+ SG_MITER_FROM_SG);

host->wait_for = MMCIF_WAIT_FOR_MWRITE;


---
base-commit: 2d5c7b7eb345249cb34d42cbc2b97b4c57ea944e
change-id: 20240220-fix-sh-mmcif-49c1de70c5b7

Best regards,
--
Linus Walleij <[email protected]>

2024-02-21 09:51:52

by Geert Uytterhoeven

Subject: Re: [PATCH] mmc: sh_mmcif: sg_miter does not need to be atomic

Hi Linus,

Thanks for your patch!

s/does not need to/must/?

On Tue, Feb 20, 2024 at 11:58 PM Linus Walleij <[email protected]> wrote:
> All the sglist iterations happen in the *threaded* interrupt handler
> and that context is not atomic, so don't request an atomic
> sglist miter. Using an atomic miter results in "BUG: scheduling while
> atomic" splats.
>
> Reported-by: Geert Uytterhoeven <[email protected]>
> Fixes: 27b57277d9ba ("mmc: sh_mmcif: Use sg_miter for PIO")
> Signed-off-by: Linus Walleij <[email protected]>
> ---
> Hi Geert, it'd be great if you could test this!

Thanks, that fixes the BUG, so
Tested-by: Geert Uytterhoeven <[email protected]>

Unfortunately it is not sufficient to make the eMMC work fully.
I will follow up in the original thread...

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68korg

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds