Hi Sergei,
On Sat, Oct 17, 2020 at 9:15 PM Sergei Shtylyov
<sergei.shtylyov@xxxxxxxxx> wrote:
> On 10/16/20 7:40 PM, Andy Shevchenko wrote:
> >> Before commit 9495b7e92f716ab2 ("driver core: platform: Initialize
> >> dma_parms for platform devices"), the R-Car SATA device didn't have DMA
> >> parameters. Hence the DMA boundary mask supplied by its driver was
> >> silently ignored, as __scsi_init_queue() doesn't check the return value
> >> of dma_set_seg_boundary(), and the default value of 0xffffffff was used.
> >>
> >> Now the device has gained DMA parameters, the driver-supplied value is
> >> used, and the following warning is printed on Salvator-XS:
> >>
> >> DMA-API: sata_rcar ee300000.sata: mapping sg segment across boundary [start=0x00000000ffffe000] [end=0x00000000ffffefff] [boundary=0x000000001ffffffe]
> >> WARNING: CPU: 5 PID: 38 at kernel/dma/debug.c:1233 debug_dma_map_sg+0x298/0x300
> >>
> >> (the range of start/end values depend on whether IOMMU support is
> >> enabled or not)
> >>
> >> The issue here is that SATA_RCAR_DMA_BOUNDARY doesn't have bit 0 set, so
> >> any typical end value, which is odd, will trigger the check.
> >>
> >> Fix this by increasing the DMA boundary value by 1.
> >>
> >> This also fixes the following WRITE DMA EXT timeout issue:
> >>
> >> # dd if=/dev/urandom of=/mnt/de1/file1-1024M bs=1M count=1024
> >> ata1.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x6 frozen
> >> ata1.00: failed command: WRITE DMA EXT
> >> ata1.00: cmd 35/00:00:00:e6:0c/00:0a:00:00:00/e0 tag 0 dma 1310720 out
> >> res 40/00:01:00:00:00/00:00:00:00:00/00 Emask 0x4 (timeout)
> >> ata1.00: status: { DRDY }
> >>
> >> as seen by Shimoda-san since commit 429120f3df2dba2b ("block: fix
> >> splitting segments on boundary masks").
> >>
> >> Fixes: 8bfbeed58665dbbf ("sata_rcar: correct 'sata_rcar_sht'")
> >> Fixes: 9495b7e92f716ab2 ("driver core: platform: Initialize dma_parms for platform devices")
> >> Fixes: 429120f3df2dba2b ("block: fix splitting segments on boundary masks")
> >> Signed-off-by: Geert Uytterhoeven <geert+renesas@xxxxxxxxx>
> >> Reviewed-by: Christoph Hellwig <hch@xxxxxx>
> >> Reviewed-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
> >> Reviewed-by: Sergei Shtylyov <sergei.shtylyov@xxxxxxxxxxxxxxxxxx>
> >> Reviewed-by: Ulf Hansson <ulf.hansson@xxxxxxxxxx>
> >> Tested-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@xxxxxxxxxxxxxx>
> >> Tested-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@xxxxxxxxxxx>
> >> Cc: stable <stable@xxxxxxxxxxxxxxx>
> >> ---
> >> v3:
> >> - Add Reviewed-by, Tested-by,
> >> - Augment description and Fixes: with Shimoda-san's problem report
> >> https://lore.kernel.org/r/1600255098-21411-1-git-send-email-yoshihiro.shimoda.uh@xxxxxxxxxxx,
> >>
> >> v2:
> >> - Add Reviewed-by, Tested-by, Cc.
> >> ---
> >> drivers/ata/sata_rcar.c | 2 +-
> >> 1 file changed, 1 insertion(+), 1 deletion(-)
> >>
> >> diff --git a/drivers/ata/sata_rcar.c b/drivers/ata/sata_rcar.c
> >> index 141ac600b64c87ef..44b0ed8f6bb8a120 100644
> >> --- a/drivers/ata/sata_rcar.c
> >> +++ b/drivers/ata/sata_rcar.c
> >> @@ -120,7 +120,7 @@
> >> /* Descriptor table word 0 bit (when DTA32M = 1) */
> >> #define SATA_RCAR_DTEND BIT(0)
> >>
> >> -#define SATA_RCAR_DMA_BOUNDARY 0x1FFFFFFEUL
> >> +#define SATA_RCAR_DMA_BOUNDARY 0x1FFFFFFFUL
> >
> > Wondering if GENMASK() here will be better to avoid such mistakes.
>
> How? The bit 0 is reserved, so only even byte counts are possiblе...
The DMA Transfer Count Register (ATAPI_DMA_TRANS_CNT) indeed does not
support odd values ("Bit 0 is ignored because the ATAPI data bus is
handled on a 16-bit basis (on a word basis)"), and is limited to
transfers less than 512 MiB.
Similarly the DMA Start Address Register (ATAPI_DMA_START_ADR) requires
that start address to be 32-bit aligned.
I believe the alignment and even size restrictions are met by the block
layer, by performing only transfers that are a multiple of the device's
block size (512 bytes minimum).
However, the SATA_RCAR_DMA_BOUNDARY definition is only used to set the
DMA boundary mask, which limits the MSB bits of an address. The LSB
bits of the mask should always be set, as a typical segment end (=
length - 1, i.e. ending in 0x1ff for a 512-byte block size) will always
have the 9 LSB bits set, else it will trigger the warning.
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
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
