There are two problems in the current timeout loop implementation: 1. In case initial test failing, there will always be a delay of 1 ms 2. The delay duration is not tunable
The new implementation addresses both these limitations.
Signed-off-by: Amit Virdi amit.virdi@st.com --- drivers/mtd/st_smi.c | 27 +++++++++++++++++---------- 1 files changed, 17 insertions(+), 10 deletions(-)
diff --git a/drivers/mtd/st_smi.c b/drivers/mtd/st_smi.c index 6d4edba..bd5602b 100644 --- a/drivers/mtd/st_smi.c +++ b/drivers/mtd/st_smi.c @@ -102,11 +102,15 @@ static struct flash_device flash_devices[] = { */ static int smi_wait_xfer_finish(int timeout) { - do { + ulong start = get_timer(0); + + while (get_timer(start) < timeout) { if (readl(&smicntl->smi_sr) & TFF) return 0; - udelay(1000); - } while (timeout--); + + /* Try after 10 ms */ + udelay(10); + };
return -1; } @@ -219,16 +223,17 @@ static int smi_read_sr(int bank) static int smi_wait_till_ready(int bank, int timeout) { int sr; + ulong start = get_timer(0);
/* One chip guarantees max 5 msec wait here after page writes, but potentially three seconds (!) after page erase. */ - do { + while (get_timer(start) < timeout) { sr = smi_read_sr(bank); if ((sr >= 0) && (!(sr & WIP_BIT))) return 0;
- /* Try again after 1m-sec */ - udelay(1000); + /* Try again after 10 usec */ + udelay(10); } while (timeout--);
printf("SMI controller is still in wait, timeout=%d\n", timeout); @@ -245,6 +250,7 @@ static int smi_wait_till_ready(int bank, int timeout) static int smi_write_enable(int bank) { u32 ctrlreg1; + u32 start; int timeout = WMODE_TOUT; int sr;
@@ -263,14 +269,15 @@ static int smi_write_enable(int bank) /* Restore the CTRL REG1 state */ writel(ctrlreg1, &smicntl->smi_cr1);
- do { + start = get_timer(0); + while (get_timer(start) < timeout) { sr = smi_read_sr(bank); if ((sr >= 0) && (sr & (1 << (bank + WM_SHIFT)))) return 0;
- /* Try again after 1m-sec */ - udelay(1000); - } while (timeout--); + /* Try again after 10 usec */ + udelay(10); + };
return -1; }