On Friday 24 February 2012 13:23:06 Amit Virdi wrote:
Curently the code makes wrong assumption that the Transfer finished flag shall be set within the stipulated time. However, there may occur a scenario in which the TFF flag is not set. Return error in that case.
Signed-off-by: Vipin Kumar vipin.kumar@st.com Signed-off-by: Amit Virdi amit.virdi@st.com
drivers/mtd/st_smi.c | 22 ++++++++++++++-------- 1 files changed, 14 insertions(+), 8 deletions(-)
diff --git a/drivers/mtd/st_smi.c b/drivers/mtd/st_smi.c index 82f1fe1..ec19b0d 100644 --- a/drivers/mtd/st_smi.c +++ b/drivers/mtd/st_smi.c @@ -58,13 +58,15 @@ static struct flash_dev flash_ids[] = {
- Wait until TFF is set in status register
*/ -static void smi_wait_xfer_finish(int timeout) +static int smi_wait_xfer_finish(int timeout) {
- while (timeout--) {
- do { if (readl(&smicntl->smi_sr) & TFF)
break;
return 0;
- }
- } while (timeout--);
- return -1;
Somewhat unrelated to the patch topic, but I don't really like this kind of timeout loops. Since it always adds at least 1ms delay after initial failing test. Better use something like this:
static int smi_wait_xfer_finish(int timeout) { ulong start = get_timer(0);
while (get_timer(start) < timeout) { if (readl(&smicntl->smi_sr) & TFF) return 0;
/* Try again after 100usec */ udelay(100); }
return -1; }
You could tune this udelay(100) down more or even remove it completely.
But such a change could be done in a addon cleanup patch, changing all timeout loops this way. I suggest you take a look at my version, here these loops are changes. In the designware networking driver as well, iirc.
Thanks, Stefan
-- DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: (+49)-8142-66989-0 Fax: (+49)-8142-66989-80 Email: office@denx.de