Hi,
the BUG was found in a Visstrim_M10 i.MX27 based board using the
following script:
#!/bin/sh
i=0
mount /dev/mmcblk0p1 /opt
while [ 0 -eq 0 ]
do
echo "Iteration $i"
cp /opt/release.trx /opt/lol.tmp
sync
MD5OLD=`md5sum /opt/release.trx | awk '{print $1}'`
MD5NEW=`md5sum /opt/lol.tmp | awk '{print $1}'`
if [ $MD5OLD = $MD5NEW ]
then
echo "OK: $MD5OLD == $MD5NEW"
else
echo "WRONG: $MD5OLD != $MD5NEW"
dmesg | tail -n 30
fi
rm -rf /opt/lol.tmp
sync
i=`expr $i + 1`
done
After several iterations the 'cp' process just hangs waiting for a MMC
transfer completion, which never happens.
We've added a timeout so that we can detect whether this problem
occurs or not. Ideally, when a MMC irq is lost we should force a retry
but it seems retries are only handled under certain circumstances by
the upper layers:
http://lxr.linux.no/#linux+v3.5.3/drivers/mmc/core/core.c#L327
Has anybody found the same problem? What is the right way to force a
retry of the data transfer in the host driver?
NOTE: the BUG only triggers when using DMA transfers, not PIO.
Regards.
--
Javier Martin
Vista Silicon S.L.
CDTUC - FASE C - Oficina S-345
Avda de los Castros s/n
39005- Santander. Cantabria. Spain
+34 942 25 32 60
www.vista-silicon.com
--
To unsubscribe from this list: send the line "unsubscribe linux-mmc" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
