write timeouts with spansion S29-WS128J

[Martin Fuzzey]

Hi,
I have an arm (i.MX21) based board with S29-WS128J flash running
something close linus's tree.

When writing to the flash I frequently see :
 MTD do_write_buffer(): software timeout

This hack stops the problem:
diff --git a/drivers/mtd/chips/cfi_cmdset_0002.c
b/drivers/mtd/chips/cfi_cmdset_0002.c

index 94bb61e..fb3cc0c 100644

--- a/drivers/mtd/chips/cfi_cmdset_0002.c

+++ b/drivers/mtd/chips/cfi_cmdset_0002.c

@@ -1312,7 +1312,7 @@ static int __xipram do_write_buffer(struct
map_info *map, struct flchip *chip,

        struct cfi_private *cfi = map->fldrv_priv;

        unsigned long timeo = jiffies + HZ;

        /* see comments in do_write_oneword() regarding uWriteTimeo. */

-       unsigned long uWriteTimeout = ( HZ / 1000 ) + 1;

+       unsigned long uWriteTimeout = ( HZ / 1000 ) + 3; /*MF avoid timeout*/

        int ret = -EIO;

        unsigned long cmd_adr;

        int z, words;



HZ is 200 on this platform.

The comments in do_write_oneword() mentionned above say that the
maximum timeout information is not available.
However if I activate DEBUG_CFI in cfi_probe.c I get:

[42949377.550000] physmap platform flash device: 04000000 at c8000000

[42949377.570000] Number of erase regions: 3

[42949377.590000] Primary Vendor Command Set: 0002 (AMD/Fujitsu Standard)

[42949377.610000] Primary Algorithm Table at 0040

[42949377.610000] Alternative Vendor Command Set: 0000 (None)

[42949377.640000] No Alternate Algorithm Table

[42949377.640000] Vcc Minimum:  1.7 V

[42949377.670000] Vcc Maximum:  1.9 V

[42949377.670000] No Vpp line

[42949377.670000] Typical byte/word write timeout: 64 µs

[42949377.710000] Maximum byte/word write timeout: 1024 µs

[42949377.710000] Typical full buffer write timeout: 512 µs

[42949377.740000] Maximum full buffer write timeout: 8192 µs
     <********************
[42949377.740000] Typical block erase timeout: 1024 ms

[42949377.770000] Maximum block erase timeout: 8192 ms

[42949377.770000] Chip erase not supported

[42949377.770000] Device size: 0x1000000 bytes (16 MiB)

[42949377.810000] Flash Device Interface description: 0x0001

[42949377.810000]   - x16-only asynchronous interface

[42949377.810000] Max. bytes in buffer write: 0x40

[42949377.850000] Number of Erase Block Regions: 3

[42949377.850000]   Erase Region #0: BlockSize 0x8000 bytes, 4 blocks

[42949377.850000]   Erase Region #1: BlockSize 0x20000 bytes, 126 blocks

[42949377.890000]   Erase Region #2: BlockSize 0x8000 bytes, 4 blocks

[42949377.890000] physmap-flash.0: Found 2 x16 devices at 0x0 in 32-bit bank

[42949377.890000] physmap-flash.0: Found 2 x16 devices at 0x2000000 in
32-bit bank

[42949377.940000]  Amd/Fujitsu Extended Query Table at 0x0040

[42949377.940000] physmap-flash.0: CFI does not contain boot bank
location. Assuming top.

[42949377.940000] number of CFI chips: 2

[42949377.980000] cfi_cmdset_0002: Disabling erase-suspend-program due
to code brokenness.



IE the max buffer write timeout is available (8192 microseconds) which
is < 1 jiffy (5ms) which probably explains the problem.
However I tried using 2 jiffies instead of 3 and it still failed sometimes.

Is this a bug or am I doing something wrong?

Regards,

Martin Fuzzey