Hi, On Mon, Jan 21, 2013 at 08:01:59PM +0200, Aaro Koskinen wrote: > Hi, > > On Mon, Jan 21, 2013 at 10:44:31AM +0200, Felipe Balbi wrote: > > On Sun, Jan 20, 2013 at 08:37:02PM +0200, Aaro Koskinen wrote: > > > Commit 0bdfe0cb803dce699ff337c35d8e97ac355fa417 (i2c: omap: sanitize > > > exit path) changed the interrupt handler to exit early and complete > > > the transfer after the draining IRQ is handled. As a result, the ARDY > > > may not be cleared properly, and it may cause all future I2C transfers > > > to timeout with "timeout waiting for bus ready". This is reproducible > > > at least with N900 when twl4030_gpio makes a long write (> FIFO size) > > > during the probe (http://marc.info/?l=linux-omap&m=135818882610432&w=2). > > > > > > The fix is to continue until we get ARDY interrupt that completes the > > > transfer. Tested with 3.8-rc4 + N900: 20 boots in a row without errors; > > > without the patch the problem triggers after few reboots. > > > > fair enough, but ... > > > > > Signed-off-by: Aaro Koskinen <aaro.koskinen@xxxxxx> > > > --- > > > drivers/i2c/busses/i2c-omap.c | 4 ++-- > > > 1 file changed, 2 insertions(+), 2 deletions(-) > > > > > > diff --git a/drivers/i2c/busses/i2c-omap.c b/drivers/i2c/busses/i2c-omap.c > > > index 832f16e..4cc2f05 100644 > > > --- a/drivers/i2c/busses/i2c-omap.c > > > +++ b/drivers/i2c/busses/i2c-omap.c > > > @@ -963,7 +963,7 @@ omap_i2c_isr_thread(int this_irq, void *dev_id) > > > i2c_omap_errata_i207(dev, stat); > > > > > > omap_i2c_ack_stat(dev, OMAP_I2C_STAT_RDR); > > > - break; > > > + continue; > > > > ... when we get draining Interrupt, it means we're receiving the last > > few bytes, meaning that by the time we handle it there's nothing pending > > to be transferred. > > > > I would rather wait for ARDY before exiting omap_i2c_xfer_msg() rather > > than here. Specially since that was the programming model suggested by > > our IP engineers. Such handling is part of my other patchset and can be > > easily ported. > > > > OTOH, we might decide to go with this during this -rc and fix it all up > > for v3.9 on my other patchset. > > Completing through ARDY interrupt has been known to work reliably for > years. So for 3.8-rc I would prefer to just restore the old behaviour. fair enough, I'll rebase my patchset on top of $SUBJECT -- balbi
Attachment:
signature.asc
Description: Digital signature
