A recent patch tried to improve the printk output of the atc_dump_lli()
function but introduced a bug, in which we end up dereferencing a
dma address as a pointer, and we even get a warning for it:
drivers/dma/at_hdmac_regs.h: In function 'atc_dump_lli':
drivers/dma/at_hdmac_regs.h:388:4: warning: format '%p' expects argument of type 'void *', but argument 3 has type 'dma_addr_t {aka long long unsigned int}' [-Wformat=]
This changes the code to pass the DMA address by reference, as expected
by printk.
Signed-off-by: Arnd Bergmann <[email protected]>
Fixes: 096117032a36 ("dmaengine: do not use 0x in front of %pad")
---
drivers/dma/at_hdmac_regs.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/dma/at_hdmac_regs.h b/drivers/dma/at_hdmac_regs.h
index 610907dbc11a..0474e4a0f02a 100644
--- a/drivers/dma/at_hdmac_regs.h
+++ b/drivers/dma/at_hdmac_regs.h
@@ -386,8 +386,8 @@ static void atc_dump_lli(struct at_dma_chan *atchan, struct at_lli *lli)
{
dev_crit(chan2dev(&atchan->chan_common),
" desc: s%pad d%pad ctrl0x%x:0x%x l%pad\n",
- lli->saddr, lli->daddr,
- lli->ctrla, lli->ctrlb, lli->dscr);
+ &lli->saddr, &lli->daddr,
+ lli->ctrla, lli->ctrlb, &lli->dscr);
}
--
2.1.0.rc2
Le 08/12/2015 16:34, Arnd Bergmann a ?crit :
> A recent patch tried to improve the printk output of the atc_dump_lli()
> function but introduced a bug, in which we end up dereferencing a
> dma address as a pointer, and we even get a warning for it:
>
> drivers/dma/at_hdmac_regs.h: In function 'atc_dump_lli':
> drivers/dma/at_hdmac_regs.h:388:4: warning: format '%p' expects argument of type 'void *', but argument 3 has type 'dma_addr_t {aka long long unsigned int}' [-Wformat=]
>
> This changes the code to pass the DMA address by reference, as expected
> by printk.
>
> Signed-off-by: Arnd Bergmann <[email protected]>
> Fixes: 096117032a36 ("dmaengine: do not use 0x in front of %pad")
Yep, thanks Arnd!
Acked-by: Nicolas Ferre <[email protected]>
> ---
> drivers/dma/at_hdmac_regs.h | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/dma/at_hdmac_regs.h b/drivers/dma/at_hdmac_regs.h
> index 610907dbc11a..0474e4a0f02a 100644
> --- a/drivers/dma/at_hdmac_regs.h
> +++ b/drivers/dma/at_hdmac_regs.h
> @@ -386,8 +386,8 @@ static void atc_dump_lli(struct at_dma_chan *atchan, struct at_lli *lli)
> {
> dev_crit(chan2dev(&atchan->chan_common),
> " desc: s%pad d%pad ctrl0x%x:0x%x l%pad\n",
> - lli->saddr, lli->daddr,
> - lli->ctrla, lli->ctrlb, lli->dscr);
> + &lli->saddr, &lli->daddr,
> + lli->ctrla, lli->ctrlb, &lli->dscr);
> }
>
>
>
--
Nicolas Ferre
On Tue, Dec 08, 2015 at 04:34:28PM +0100, Arnd Bergmann wrote:
> A recent patch tried to improve the printk output of the atc_dump_lli()
> function but introduced a bug, in which we end up dereferencing a
> dma address as a pointer, and we even get a warning for it:
>
> drivers/dma/at_hdmac_regs.h: In function 'atc_dump_lli':
> drivers/dma/at_hdmac_regs.h:388:4: warning: format '%p' expects argument of type 'void *', but argument 3 has type 'dma_addr_t {aka long long unsigned int}' [-Wformat=]
>
> This changes the code to pass the DMA address by reference, as expected
> by printk.
Thanks for this but i was planning to drop 096117032a36 "dmaengine: do not
use 0x in front of %pad" as recomended by Dmitry, so we should fix this
cleanly
I will push out updated tree in a short while...
>
> Signed-off-by: Arnd Bergmann <[email protected]>
> Fixes: 096117032a36 ("dmaengine: do not use 0x in front of %pad")
> ---
> drivers/dma/at_hdmac_regs.h | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/dma/at_hdmac_regs.h b/drivers/dma/at_hdmac_regs.h
> index 610907dbc11a..0474e4a0f02a 100644
> --- a/drivers/dma/at_hdmac_regs.h
> +++ b/drivers/dma/at_hdmac_regs.h
> @@ -386,8 +386,8 @@ static void atc_dump_lli(struct at_dma_chan *atchan, struct at_lli *lli)
> {
> dev_crit(chan2dev(&atchan->chan_common),
> " desc: s%pad d%pad ctrl0x%x:0x%x l%pad\n",
> - lli->saddr, lli->daddr,
> - lli->ctrla, lli->ctrlb, lli->dscr);
> + &lli->saddr, &lli->daddr,
> + lli->ctrla, lli->ctrlb, &lli->dscr);
> }
>
>
> --
> 2.1.0.rc2
>
>
--
~Vinod
>> A recent patch tried to improve the printk output of the atc_dump_lli()
>> function but introduced a bug
BTW, my original patch was created for mainline and was incorrectly
applied to slave-dma.git for-linus branch (I sent Vinod an email about
that).
In that branch the code was already changed so that my patch didn't
make sense there.
Now you seem to revert the change that was originally made instead of
revering my 1 line change:) Is that intentional?
Thanks,
Dmitry