| Message ID | 1425320166-13788-3-git-send-email-andriy.shevchenko@linux.intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Mon, Mar 02, 2015 at 08:15:59PM +0200, Andy Shevchenko wrote: > The logic of DMA completion is broken now since test_and_clear_bit() never > returns the other bit is set. It means condition are always false and we have > spi_finalize_current_transfer() called per each DMA completion which is wrong. > > The patch fixes logic by clearing BUSY bit first and then check for the other > one. > > Fixes: 30c8eb52cc4a (spi: dw-mid: split rx and tx callbacks when DMA) The commit this is fixing is in v3.19 so this should go to stable but I can't apply this against my fix/dw branch. Can you generate a version that can be applied there please? We'll need to work out the merge issues with the current code I gues...
On Fri, 2015-03-06 at 11:25 +0000, Mark Brown wrote: > On Mon, Mar 02, 2015 at 08:15:59PM +0200, Andy Shevchenko wrote: > > The logic of DMA completion is broken now since test_and_clear_bit() never > > returns the other bit is set. It means condition are always false and we have > > spi_finalize_current_transfer() called per each DMA completion which is wrong. > > > > The patch fixes logic by clearing BUSY bit first and then check for the other > > one. > > > > Fixes: 30c8eb52cc4a (spi: dw-mid: split rx and tx callbacks when DMA) > > The commit this is fixing is in v3.19 so this should go to stable but I > can't apply this against my fix/dw branch. Can you generate a version > that can be applied there please? We'll need to work out the merge > issues with the current code I gues... Hmm… Just tried apply on top on fix/dw using git am -C1 -s 0001-spi-dw-mid-clear-BUSY-flag-fist-and-test-other-one.patch It went fine with reduced context. Shall I still resend you that version?
On Fri, Mar 06, 2015 at 02:40:29PM +0200, Andy Shevchenko wrote: > On Fri, 2015-03-06 at 11:25 +0000, Mark Brown wrote: > > The commit this is fixing is in v3.19 so this should go to stable but I > > can't apply this against my fix/dw branch. Can you generate a version > > that can be applied there please? We'll need to work out the merge > > issues with the current code I gues... > Hmm… Just tried apply on top on fix/dw using > git am -C1 -s > 0001-spi-dw-mid-clear-BUSY-flag-fist-and-test-other-one.patch > It went fine with reduced context. > Shall I still resend you that version? Yes, please resend.
diff --git a/drivers/spi/spi-dw-mid.c b/drivers/spi/spi-dw-mid.c index a8a5ee3..c8416ef 100644 --- a/drivers/spi/spi-dw-mid.c +++ b/drivers/spi/spi-dw-mid.c @@ -108,7 +108,8 @@ static void dw_spi_dma_tx_done(void *arg) { struct dw_spi *dws = arg; - if (test_and_clear_bit(TX_BUSY, &dws->dma_chan_busy) & BIT(RX_BUSY)) + clear_bit(TX_BUSY, &dws->dma_chan_busy); + if (test_bit(RX_BUSY, &dws->dma_chan_busy)) return; spi_finalize_current_transfer(dws->master); } @@ -156,7 +157,8 @@ static void dw_spi_dma_rx_done(void *arg) { struct dw_spi *dws = arg; - if (test_and_clear_bit(RX_BUSY, &dws->dma_chan_busy) & BIT(TX_BUSY)) + clear_bit(RX_BUSY, &dws->dma_chan_busy); + if (test_bit(TX_BUSY, &dws->dma_chan_busy)) return; spi_finalize_current_transfer(dws->master); }
The logic of DMA completion is broken now since test_and_clear_bit() never returns the other bit is set. It means condition are always false and we have spi_finalize_current_transfer() called per each DMA completion which is wrong. The patch fixes logic by clearing BUSY bit first and then check for the other one. Fixes: 30c8eb52cc4a (spi: dw-mid: split rx and tx callbacks when DMA) Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> --- drivers/spi/spi-dw-mid.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)