Opened 12 years ago
Closed 12 years ago
#48 closed defect (duplicate)
freeze writing JFS memory stick (version 189)
Reported by: | Barry Landy | Owned by: | somebody |
---|---|---|---|
Priority: | major | Component: | basedrv |
Version: | Keywords: | ||
Cc: | bl10@… |
Description
Same problem as reported earlier for previous versions of the suite.
While writing a lot of data (zip files of various sizes) to a JFS formatted partition on a memory stick the entire system freezes. The USB widget generally reports an invalid removal at the same time.
The likelihood of this is increased by other things (non USB) happening in parallel and reduced 9but not to zero) but doing NOTHING else in parallel.
Change History (11)
comment:1 by , 12 years ago
comment:2 by , 12 years ago
Will do. On previous versions this was suggested but did not turn out to be the culprit.
PS: for some reason the email link seems to have vanished.
comment:3 by , 12 years ago
Cc: | added |
---|
comment:4 by , 12 years ago
Tried without the widget and three writes of 90 zips to the JFS mem stick worked OK, the last with me running lots of things from another session including simultaneous access to a second partition on the same stick.
Will keep using it and report back if trouble.
comment:5 by , 12 years ago
I have now been able to provoke a freeze in version 189 without the widget present by doing the multi JFS write to mem stick and doing other things in parallel.
comment:6 by , 12 years ago
There are 2 points to consider:
1) JFS is not meant to be used with a removable device. However, if you have attached a USB mass memory device already on bootup it will be handled as a fixed disk (the drive icon will indicate that) and then it should be ok to use JFS. I have a WD book permanently attached. It's JFS formatted and works ok. However I don't do large data transfers to it.
2) the error is not in the HC drivers, see here:
http://newsgroups.derkeiler.com/Archive/Comp/comp.os.os2.misc/2012-10/msg00086.html
and follow ups. In short: it's a problem either in USBMSD (more likely) or in OS2DASD (less likely). I suspect it now occurs more frequently because the transfer performance of USBEHCD has increased. That will lead to more IORBs being serviced and maybe USBMSD is missing some boundary to limit and properly serialize the requests. If it's a problem with too many transfers, you can use the /S: switch on USBEHCD.SYS to reduce the throughput.
3) I am not sure if it is related: occasionally I have complete system freezes with Seamonkey even though I am not doing anything with USB at the moment of the freeze. It freezes when I enter a search term on the google page and google auto completes my entry. Therefore: as a test, do not run Firefox/Thunderbird/Seamonkey when you do large data transfers to your JFS volume. See if that makes a difference.
comment:7 by , 12 years ago
OK then.
1) it is no longer giving a trap 8 though it might be the same effect
2) I agree it is more likely to work when nothing else is running; the one thing I sometimes leave running is Firefox running google calender and then it is indeed more likely to freeze,
so...
3) I will revert to using HPFS on the stick and expect that will not give problems.
Incidentally why do you say JFS is not intended to be used on a removable?
comment:8 by , 12 years ago
Trying with the stick formatted HPFS and not worrying what else I was doing, I got a forced reboot.....
I will try in isolated mode (and also FAT32)
Not sure what diagnostics I could usefully produce
comment:9 by , 12 years ago
More details. One mem stick; two partitions. One is formatted FAT32, the other HPFS.
Writing to the FAT32 partition seems to work every time.
Writing to the HPFS partition seems to always stick very quickly (except for the one time it caused a reboot). in this state I can recover the system (but not the writes) but removing the stick from the usb slot.
This may be the same problem as dbisse. (#45 i think)
comment:10 by , 12 years ago
If it is of interest.
On my laptop (T61) I am using version 188.
Writing to the HPFS formatted partition on the same memory stick caused a trap which I have captured in a dump.
Would you like to see the dump?
Try without the USB widget. See readme.txt.