summaryrefslogtreecommitdiffstats
path: root/fs/pnode.c
diff options
context:
space:
mode:
authorQipan Li <Qipan.Li@csr.com>2013-05-18 19:46:06 +0800
committerMark Brown <broonie@opensource.wolfsonmicro.com>2013-05-20 11:10:49 -0500
commit237ce4665c12376391ddb148509204652f6ab758 (patch)
treeef66edad9f6fecec1082b75d62ea1c150ea84ff0 /fs/pnode.c
parentde1f9f270ea7fb7af37e1b62580e27b15273d63d (diff)
downloadop-kernel-dev-237ce4665c12376391ddb148509204652f6ab758.zip
op-kernel-dev-237ce4665c12376391ddb148509204652f6ab758.tar.gz
spi: sirf: fix the issue while transferring more than 256 words
currently, spi irq handler only does rx processing and fetching data from rx fifo when "FRM_END" irq happens. FRM_END indicates one transfer completes. if rx size is less than 256, it works well. but the problem is that spi rx fifo size is only 256 bytes, then if data size of one frame is more than 256, before FRM_END comes, rx fifo will be filled with RXFIFO_OFLOW overflow interrupt, it will make us lose some data due to fifo overflow. Explicitly we need do fetch work from device rx fifo in irq handler not only in "FRM_END" irq but also in "THD_REACH" irq. THD_REACH means rx fifo has come to its threshold and will come to overflow if we don't take data from it in time. In this patch, we fix this issue. we take data from rx fifo when either FRM_END or RX_THD_REACH irq comes, we put data into tx fifo when either TX_FIFO_EMPTY or TX_THD_REACH irq comes. Signed-off-by: Qipan Li <Qipan.Li@csr.com> Signed-off-by: Zhiwu Song <Zhiwu.Song@csr.com> Signed-off-by: Barry Song <Baohua.Song@csr.com> Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Diffstat (limited to 'fs/pnode.c')
0 files changed, 0 insertions, 0 deletions
OpenPOWER on IntegriCloud