View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0001255 | channel: elrepo/el9 | kmod-hfsplus | public | 2022-08-25 15:47 | 2023-07-04 23:03 |
| Reporter | rfs34 | Assigned To | toracat | ||
| Priority | normal | Severity | major | Reproducibility | always |
| Status | resolved | Resolution | fixed | ||
| Platform | x86_64 | OS | rockylinux | OS Version | 9.0 |
| Summary | 0001255: Given two memory sticks both containing hfs-plus formatted partitions, cp -a * from one to the other always causes crash+reboot | ||||
| Description | Packages in use: kmod-hfsplus-0.2-1.el9_0.elrepo.x86_64.rpm hfsplus-tools-540.1.linux3-26.el9.x86_64.rpm How reproducible: Has happened on 5 occasions out of 5. Steps to produce problem: Format a memory stick partition using mkfs.hfsplus, mount it (say on <mnt1>), and put in it the contents of the hfsplus partition from a standard rockylinux 9.0 livestick (let me know if you'd like me to send you those contents, but I suspect that all that matters is that all the files are owned by root apart from <mnt1>/.. which is owned by somebody else, and that they have various different permissions - the output of ls -laR <mnt1> is below). Take a second memory stick, again format one partition using mkfs.hfsplus, mount it (say on <mnt2>). Run cp -a <mnt1>/* <mnt2> Result: crash and reboot. The problem did *not* happen if I omitted the -a argument and instead gave -r. Nor did it happen if copying from a hard disk partition to a memory stick partition or vice versa. | ||||
| Additional Information | Problem persists after reboot, except sometimes the partition mounted on <mnt2> becomes unwriteable. | ||||
| Tags | No tags attached. | ||||
|
|
The current kmod package was built from the source code in the RHEL 9.0 kernel. This is based on kernel-5.14. What we could try is to rebuild it by using the code from a newer kernel. |
|
|
The kernel version in which the bug was observed was 5.14.0-70.17.1.el9_0.x86_64, in case it makes a difference. |
|
|
I rebuilt the kmod using the code from the latest kernel, 5.19.4. This failed to build. Then I tried the most recent stable kernel, 5.15.63. The build was successful. kmod-hfsplus-0.4-1.el9_0.elrepo.x86_64.rpm is now syncing to the mirror sites. |
|
|
Thank you toracat. Sadly your latest version kmod-hfsplus-0.4-1.el9_0.elrepo.x86_64.rpm gives exactly the same whole system crash. I also tried taking the source of this latest version and compiling it against my actual current kernel, and that gave exactly the same whole system crash. |
|
|
I notice that blkdev_issue_flush used to need an extra parameter GFP_KERNEL in the calls to it in inode.c and super.c, but that it is no longer capable of taking this extra argument, although a lot of other functions called still need it. Is this relevant ? (I have absolutely no idea what these functions do or what the meaning of this parameter is.) |
|
|
Since 5.19.4 is an LTS kernel and is still supported, I wonder if this is something you can report as a bug upstream (kernel.org). Yes, it would be nice if we can find the cause and provide a patch. But that is beyond what ELRepo can offer. |
|
|
Thank you Toracat - please forgive my ignorance, I had thought that elrepo was the originating site for this. I will report it to kernel.org as you suggest. |
|
|
Just a heads up. RHEL 9.1 is out. Building the kmod-hfsplus package on 9.1 results in errors. Will try to fix it but no ETA. |
|
|
Updated packages for el9.1 have been released to the testing repository: kmod-hfsplus-0.5-1.el9_1.elrepo.x86_64.rpm Please could you test (on el9.1 once available to you) and confirm the updated package works as expected for you. This package will not likely support / be compatible with el9.0. |
|
|
Thank you Phil and toracat for your efforts on this. Please expect this in second or third week of December - I need rocky 9.1 to be available and term to have ended before I can afford to spend time upgrading. |
|
|
No worries, we will appreciate the feedback whenever you are able, and fully understand the constraints on your valuable time. Do let us know how you get on and we will be happy to support in any way we are able should you have any issues. |
|
|
Phil, Akemi, Thank you for getting this kmod to work with 9.1; I now confirm that it works as expected - i.e. everything works except that the bug reported is still present (but I understand that that is an upstream matter and not your concern). Thanks again for your assistance. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2022-08-25 15:47 | rfs34 | New Issue | |
| 2022-08-25 16:41 | pperry | Assigned To | => toracat |
| 2022-08-25 16:41 | pperry | Status | new => assigned |
| 2022-08-26 13:27 | toracat | Note Added: 0008545 | |
| 2022-08-26 15:42 | rfs34 | Note Added: 0008546 | |
| 2022-08-27 04:42 | toracat | Note Added: 0008547 | |
| 2022-08-27 08:04 | rfs34 | Note Added: 0008548 | |
| 2022-08-27 08:09 | rfs34 | Note Added: 0008549 | |
| 2022-08-27 14:31 | toracat | Note Added: 0008550 | |
| 2022-08-27 15:33 | rfs34 | Note Added: 0008551 | |
| 2022-11-16 13:15 | toracat | Note Added: 0008748 | |
| 2022-11-17 03:52 | pperry | Note Added: 0008749 | |
| 2022-11-17 03:52 | pperry | Status | assigned => feedback |
| 2022-11-17 04:32 | rfs34 | Note Added: 0008751 | |
| 2022-11-17 04:32 | rfs34 | Status | feedback => assigned |
| 2022-11-17 08:10 | pperry | Note Added: 0008752 | |
| 2022-12-07 08:27 | rfs34 | Note Added: 0008829 | |
| 2023-07-04 23:03 | toracat | Status | assigned => resolved |
| 2023-07-04 23:03 | toracat | Resolution | open => fixed |
