KDL when attempting to burn CD

[vidrep]

hrev53365 x86_64

Attempting to burn a CD caused a KDL.

Attached syslog and photo.

[waddlesplash]

It appears cdrecord is actually using BeOS areas internally for operation. They now need to have the B_CLONEABLE_AREA flag applied; please open a bug at HaikuPorts.

[korli]

IMO binaries built for r1b1 shouldn't be affected by the B_CLONEABLE_AREA flag policy change of a public API. It should be possible to fix in Haiku (if r1b1 binary => use always B_CLONEABLE_AREA ).

[waddlesplash]

Writing and then applying some sort of heuristic to detect that would be pretty tricky, I don't know what you could look at to determine that. But even if there was, is it worth the time and effort to fix one application? If there were a lot more that might make sense, but so far no others have been reported.

[korli]

One can check image.api_version < B_HAIKU_VERSION_1_PRE_BETA_2.

[waddlesplash]

That would require fetching the image of the return address before every create_area, which would complicate the code significantly; or in the case where we used __gABIVersion, it would pick the oldest version, which would always be a pre-beta2 image, making the clone_area changes useless.

I still maintain that if cdrecord is the only affected application, we should just add some patch to it (or even just wait for the beta2 branch cut) and move on.

[vidrep]

I tried to get a little more info.When attempting to burn a cd, cdrecord crashes. After that, it is impossible to eject the disc unless cdrecord -eject is invoked from the CLI. I attached debugger to cdrecord in this case. Debug reports attached.

[korli]

see ​https://review.haiku-os.org/c/haiku/+/1849

[waddlesplash]

Change merged in hrev53495.

[nielx]

Assign tickets with status=closed and resolution=fixed within the R1/beta2 development window to the R1/beta2 Milestone