disk test failure when run "Random data with random seeking test pattern "

[eric.lai]

Hi.
There is a eMMC on my board but has two SKU one is 16GB another one is 32GB. (OS: WIN10, Burn in V8.1 1020 build)
Using eMMC as local disk and run burn in test with "Random data with random seeking test pattern", the 16GB eMMC has no porblem but 32GB eMMC always fail in "write to disk".
The setting of block size: 4MB, seek count: 100.

I try to change the block size to 128K and 64M when test with 32GB eMMC, then the problem does happen anymore.
I'm confusing about the test result, due to the configurations are all the same just the dfferent capacity.
Do you have any experience of this problem.

Thanks!

[Tim (PassMark)]

If you set the trace file detail level to activity trace 2 in the logging options then some more details about the error should be logged.

[eric.lai]

Hi,
The trace log shows as below when the error happens and attached log file for you reference. (plase modify the file type of BIT_log_ to .rar and decompression)
Could you assist me in tracing log analysis?
Thanks a lot~

================================================== ================================================== =====
LOG NOTE: 2017-01-15 08:32:30, Disk , Seeking distance 134217728, to block number 33 within file D:\BurnInTest test files\~bittestD00012
LOG NOTE: 2017-01-15 08:32:30, Disk , [D:] Writing block size 4194304
LOG NOTE: 2017-01-15 08:32:30, Disk , Mem Tot (D2): 3945MB [2980MB], page: 4649MB [3673MB], virt: 134217728MB [134217440MB], ext virt: [0MB], 24
LOG NOTE: 2017-01-15 08:32:30, Disk , Disk free space: 1560MB, disk size: 28029MB
LOG NOTE: 2017-01-15 08:32:30, Disk , Mem Tot (D3): 3945MB [2980MB], page: 4649MB [3673MB], virt: 134217728MB [134217440MB], ext virt: [0MB], 24
LOG NOTE: 2017-01-15 08:32:30, Disk , Mem Tot (D1): 3945MB [2980MB], page: 4649MB [3673MB], virt: 134217728MB [134217440MB], ext virt: [0MB], 24
LOG NOTE: 2017-01-15 08:32:30, Disk , Writing file: D:\BurnInTest test files\~bittestD00063, size: 293601280, block size: 4194304
LOG NOTE: 2017-01-15 08:32:30, Disk , Mem Tot (W1): 3945MB [2980MB], page: 4649MB [3673MB], virt: 134217728MB [134217440MB], ext virt: [0MB], 24
LOG NOTE: 2017-01-15 08:32:30, Disk , WriteFile - creating file: D:\BurnInTest test files\~bittestD00063
LOG NOTE: 2017-01-15 08:32:30, Disk , Mem Tot (W2): 3945MB [2980MB], page: 4649MB [3673MB], virt: 134217728MB [134217440MB], ext virt: [0MB], 24
LOG NOTE: 2017-01-15 08:32:30, Disk , WriteFile - write file header: D:\BurnInTest test files\~bittestD00063
LOG NOTE: 2017-01-15 08:32:30, Disk , Mem Tot (W3): 3945MB [2980MB], page: 4649MB [3673MB], virt: 134217728MB [134217440MB], ext virt: [0MB], 24
LOG NOTE: 2017-01-15 08:32:30, Disk , Mem Tot (W4): 3945MB [2980MB], page: 4649MB [3673MB], virt: 134217728MB [134217440MB], ext virt: [0MB], 24
LOG NOTE: 2017-01-15 08:32:30, Disk , WriteFile - write file remainder: D:\BurnInTest test files\~bittestD00063
LOG NOTE: 2017-01-15 08:32:30, Disk , Mem Tot (W5): 3945MB [2980MB], page: 4649MB [3673MB], virt: 134217728MB [134217440MB], ext virt: [0MB], 24
LOG NOTE: 2017-01-15 08:33:13, Disk , WriteFile error : D:\BurnInTest test files\~bittestD00063, block 2064d (121)
SERIOUS: 2017-01-15 08:33:13, Disk, D: Error while writing to the disk
LOG NOTE: 2017-01-15 08:33:13, Disk, Write File n: D:\BurnInTest test files\~bittestD00063, Block Size: 4194304, Block Num: 21, Ops: 264828354560 (B)
LOG NOTE: 2017-01-15 08:33:13, Disk, The semaphore timeout period has expired. (121)

[Tim (PassMark)]

Given the error message (121 = ERROR_SEM_TIMEOUT The semaphore timeout period has expired) it sounds like there is an internal buffer filling up faster than can be written to the device and the device driver isn't able to handle it, there is a resource leak in the device driver or the device driver has crashed.

This is not a very common error to see when writing to a storage device and is more often associated with network communications (eg copying files across a network) so there is likely an issue with the device driver for the eMMC controller.

[eric.lai]

I ran test with bloack size 128K/4M/64M, only error in 4M.
Are there any relationship with block size?

[Tim (PassMark)]

This particular block size may be triggering a specific condition in the device driver to cause the behaviour.