Ticket #2699 (closed bug: fixed)

Opened 4 months ago

Last modified 4 months ago

KDL when accessing a DVD or CD

Reported by: cebif Owned by: axeld
Priority: critical Milestone: R1/alpha1
Component: File Systems/ISO 9660 Version: R1 development
Cc: Blocked By:
Platform: All Blocking:

Description

I mounted a DVD movie and double clicked on its icon. This resulted in KDL. I then tried right clicking. The same thing again. I tried to open it from VLC, via the file menu but KDL again. I then tried a CDR with Mandriva Linux burned on it (iso9660), KDL again. To reproduce: 1. Mount a DVD or CD disk. 2. Double click or right click or navigate to from VLC or MediaPlayer, to open. The result is immediate KDL. I will upload a sc from KDL when I tried to open the DVD movie.

Attachments

SC from KDL opening DVD movie.jpg (486.6 KB) - added by cebif 4 months ago.
Stacktrace from KDL double clicking on DVD movie icon.

Change History

Changed 4 months ago by cebif

Stacktrace from KDL double clicking on DVD movie icon.

  Changed 4 months ago by cebif

I was testing r27258.

  Changed 4 months ago by emitrax

Just to make sure, did you add udf to the iso? Because by default it is not in the image yet.

follow-up: ↓ 4   Changed 4 months ago by emitrax

Sorry, by iso I mean tha haiku image.

in reply to: ↑ 3   Changed 4 months ago by cebif

Replying to emitrax:

Sorry, by iso I mean tha haiku image.

No, I did not compile a haiku image. The crash happens with different types of file systems but I have only tested with iso9660, DVD movie and a DVD+RW I burned but not certain of its fs. Is udf in DVD movie? I will have to check what the different fs types are on the disks.

  Changed 4 months ago by cebif

Found the DVD+RW is burned with iso9660.

follow-up: ↓ 7   Changed 4 months ago by emitrax

As far as I can tell from the bt this does not seem to be related to the FS. It seems to be a tracker problem, and IIRC I traced it once with Renè. The ASSERT that triggers the panic has been introduced by Axel in r27128, and it is involved for user space caller only. What I don't understand it is why the ASSERT is not trigger everytime you try to access a media, as the code path involved should be the same. Perhaps someone else can tell you (and me) more.

in reply to: ↑ 6   Changed 4 months ago by cebif

Replying to emitrax:

As far as I can tell from the bt this does not seem to be related to the FS. It seems to be a tracker problem, and IIRC I traced it once with Renè. The ASSERT that triggers the panic has been introduced by Axel in r27128, and it is involved for user space caller only. What I don't understand it is why the ASSERT is not trigger everytime you try to access a media, as the code path involved should be the same. Perhaps someone else can tell you (and me) more.

When you say "everytime you try to access media", do you mean any CD's or other disks with video or audio? I tried to access a data CD and a data DVD and the crash happened as well. Or do you mean all types of physical removeable media?

  Changed 4 months ago by cebif

I have tested with a USB digital camera SD memory and a audio CDROM. I can make the crash happen when the SD memory is mounted in read/write but not read-only. I cannot make the crash happen with the audio CD. The audio CD was is off course read-only but a choice is given to mount read/write or read-only when mounting. I don't know what would happen if I selected read/write.

  Changed 4 months ago by cebif

I can make the crash happen for all removeable disks except audio CDROMS. With a USB SD memory card; the crash will only happen if it is mounted read/write.

  Changed 4 months ago by kev

I'm having this problem as well, trying to read a DVD-R that I burned in UDF102 format in Windows XP using Roxio Creator Classic. Instant KDL when I double-click the desktop icon (even though it auto-mounts fine.) This is on r27393.

  Changed 4 months ago by kev

I guess the build factory is still down...any word on how close udf is to being included by default?

  Changed 4 months ago by emitrax

This has nothing to do with UDF. It's more likely a tracker problem, as I sais already. Have you tried listing the content of the media from terminal?

As for udf being included to the image by default, there is bug that have to be fixed first.

  Changed 4 months ago by axeld

  • priority changed from normal to critical
  • resolution set to fixed
  • status changed from new to closed
  • component changed from System/libroot.so to File Systems/ISO 9660
  • milestone changed from R1 to R1/alpha1

Fixed in r27408.

  Changed 4 months ago by cebif

I am still getting the bug here, testing with r27407 installed on HD. Most of the stacktraces are the same except for one, when I tried to open a mounted data DVD. Can I open this again?

  Changed 4 months ago by cebif

Sorry I didn't notice I was testing one version behind r27408. It was the latest version available so I thought I had downloaded an image with the fix.

follow-up: ↓ 17   Changed 4 months ago by cebif

Finally made sure to get the right image r27420. There are no more crashes but any CD or DVD I mount and open show as blank but they have folders and files. It doesn't matter if they have been burned or are bought disks. Is this part of the same bug or another one?

in reply to: ↑ 16 ; follow-up: ↓ 18   Changed 4 months ago by emitrax

Replying to cebif:

Finally made sure to get the right image r27420. There are no more crashes but any CD or DVD I mount and open show as blank but they have folders and files. It doesn't matter if they have been burned or are bought disks. Is this part of the same bug or another one?

If the media are UDF formatted, that's unfortunately the expected behavior for now, since there is not udf module and they get picked by the iso9660 one.

in reply to: ↑ 17   Changed 4 months ago by cebif

Replying to emitrax:

Replying to cebif:

Finally made sure to get the right image r27420. There are no more crashes but any CD or DVD I mount and open show as blank but they have folders and files. It doesn't matter if they have been burned or are bought disks. Is this part of the same bug or another one?

If the media are UDF formatted, that's unfortunately the expected behavior for now, since there is not udf module and they get picked by the iso9660 one.

They are not UDF formatted. They were mostly iso9660.

  Changed 4 months ago by cebif

The only two type CD's I can veiw the contents of is an old BeOSR4.5.2 install disk. This part of the disk is formatted in bfs. It also has another partition on it that runs from Windows for creating a HD partition to intall Beos. That CD partition, I cannot view the contents of. It is formatted in iso9660. I can also view the contents of an audio CD. It is formatted in cdda.

  Changed 4 months ago by cebif

  • status changed from closed to reopened
  • resolution fixed deleted

I just tried mounting a digital camera SD memory card in read write mode and got KDL when I double clicked on its icon. I did a stacktrace and it is similar to the original stacktrace for this bug. If I open it in read only there is no crash. I think I will reopen this.

  Changed 4 months ago by cebif

The stacktrace for Ticket #2700 looks the same as the trace for this bug.

follow-up: ↓ 23   Changed 4 months ago by axeld

  • status changed from reopened to closed
  • resolution set to fixed

An SD card has nothing to do with the ISO9660 file system, please don't just reopen tickets where you see fit.

If you have a file system different to ISO9660 (this bug), or ext3 (bug #2700), then please just open a new ticket.

in reply to: ↑ 22   Changed 4 months ago by cebif

Replying to axeld:

An SD card has nothing to do with the ISO9660 file system, please don't just reopen tickets where you see fit. If you have a file system different to ISO9660 (this bug), or ext3 (bug #2700), then please just open a new ticket.

I thought it was more or less clear this bug was not reported as just applying to iso9660. I also thought reopening a bug was allowable to non developers otherwise the radio box would not have been available to me. Just like it is not available to a non developer to delete an invalid backtrace or attachment.

follow-up: ↓ 25   Changed 4 months ago by axeld

Have you read the summary or the description? It clearly mentions CDs and DVDs as well as iso9660, and nothing more. How could it be clear that this bug was about everything else?

While it's technically allowed for you to reopen bugs, it's just annoying when you do so incorrectly. When in doubt, just ask before. This was an iso9660 specific bug, and it has been fixed. This is clearly visible by the component this bug has been assigned to as well as the changeset that fixed it.

If the CD appears empty, than this is just another bug in iso9660.

If your SD card let's you drop into KDL, this is not a problem of iso9660.

in reply to: ↑ 24   Changed 4 months ago by cebif

Replying to axeld:

Have you read the summary or the description? It clearly mentions CDs and DVDs as well as iso9660, and nothing more. How could it be clear that this bug was about everything else? While it's technically allowed for you to reopen bugs, it's just annoying when you do so incorrectly. When in doubt, just ask before. This was an iso9660 specific bug, and it has been fixed. This is clearly visible by the component this bug has been assigned to as well as the changeset that fixed it. If the CD appears empty, than this is just another bug in iso9660. If your SD card let's you drop into KDL, this is not a problem of iso9660.

Sorry for feeling a bit hot here. I'm the one that reported this bug I should know what I stated and meant. I know I stated iso9660 as one of the file systems but I wasn't saying that it was a file system specific bug. At that point I wasn't certain. I was wanting to test more as shown in a slightly later comment to emitrax:

"Replying to emitrax:

Sorry, by iso I mean tha haiku image.

No, I did not compile a haiku image. The crash happens with different types of file systems but I have only tested with iso9660, DVD movie and a DVD+RW I burned but not certain of its fs. Is udf in DVD movie? I will have to check what the different fs types are on the disks." Although I didn't say it outright I was more under the impression it was a tracker type bug, as stated outright by emitrax. In my initial report I had the component as: libroot.so. I'm not saying you should have left it at that (I know you're knowledge is greater) but that shows I wasn't saying the component was iso9660. Yes I probably should have asked again. I did ask once but this whole comment (that I asked it in can be looked at as invalid because I had the wrong image). Sorry for not doing that. I was feeling frustrated because one other bug I reported before this was almost immediately marked as copy (not by you) when there was crucial difference. I asked to reopen it and the reason why but have not had a reply in weeks. Any way that particular bug is really fixed now. Is my bug or any one elses vanity.

Note: See TracTickets for help on using tickets.