November 2000 - Packet Writing - openSUSE Mailing Lists

freeze with0.0.2d
by Mark Hindley 24 Nov '00

24 Nov '00

I ahve kernel 2.4.0-test10 patched with 0.0.2d The pktcdvd module loads fine. I can mount a UDF formatted CDRW, but if I try to write to it I get a total freeze of the system. A hard reset is the only way out. Yamaha 2214 IDE scsi-ide emulation The logs are empty. Does anybody else have problems? thanks Mark

3 3

Kernel locks upon write
by Daniel Tschan 22 Nov '00

22 Nov '00

Hi I just installed the packet-writing patch and the newest udf utilities (from CVS). I am using kernel 2.4.0-test10 and a HP CD-Writer Plus 9200i (SCSI). I did everything written in the installation instructions, no problem there. But whenever I try to write something on a CD-RW the kernel locks immediately. All I can do then is pressing the reset button. Reading works without problems. I saw that the writing speed is detected incorrectly (8x instead of 4x). Maybe this is the reason for the kernel locks. Is there a possibility to override this with an option in /etc/modules.conf? Daniel Tschan

1 0

A reading TOC bug?
by jack 09 Nov '00

09 Nov '00

Dear sir, Sorry for bothering you again. I have send a mail to you several days ago. But you have not replied me yet. I am afraid that mail was lost or something.... So, I decide to write a mail to you again. I found the uniform cdrom driver seem to hide a bug on reading the TOC of a disc. I think the start address of the last track of the last session that has more than one track is always wrong. I have examined this repeating several times by CDpro (a utility under windows) and my VCD play program. The running kernel version on my PC is 2.2.17. Is this a known bug? Or I make any mistake? By the way, is there any easy way to soft reset cdrom? Thank you for your help! Sincerely, Jack Ciao Nov 9

1 0

Lost interrupt error in cdrwtool
by Marco Cavalieri 07 Nov '00

07 Nov '00

Hi. I own an IDE CD/RW writer (Waitec SHUTTLE) and I am experiencing a problem with packet0.0.2d. When I try to initialize a CD with: cdrwtool -d /dev/hdd -q -t 4 the CD writer starts writing (orange light on). After about 15-20 minutes, the writing phase ends, but the writer is in error condition (green light flashing). In this situation, the cdrwtool process is hanging (no response to Ctl-C nor kill -9), the CD tray is locked and the system console repeatedly says "hdd: lost interrupt". This is the relevant part of syslog: Nov 4 15:05:08 star510 kernel: hdd: packet command error: status=0x51 DriveReady SeekComplete Error } Nov 4 15:05:08 star510 kernel: hdd: packet command error: error=0x50 Nov 4 15:05:08 star510 kernel: cdrom: This disc doesn't have any tracks I recognize! Nov 4 15:23:01 star510 /USR/SBIN/CRON[16759]: (mail) CMD (runq) Nov 4 15:43:01 star510 /USR/SBIN/CRON[16763]: (mail) CMD (runq) Nov 4 15:47:57 star510 kernel: hdd: lost interrupt Nov 4 15:48:37 star510 last message repeated 4 times Nov 4 15:49:47 star510 last message repeated 7 times Can someone help me? Does packet-cd require some feature that my hardware does not provide?

1 0

A reading TOC bug?
by jack 06 Nov '00

06 Nov '00

Hello, I found that the start addresses of the last track are different between CDROMREADENTRY and CDROM_PACKET_COMMAND (with parameter [43 02 02 00 00 00 00 10 00 00 00 00]). For example, there is a VCD disc that has one session and 12 tracks. I use a widows utility "CDpro" to observe this disc. CDpro shows the 12th track is start at LBN 271601 and end at 272465. I use ioctl with CDROMREADENTRY will get the same start address. But, when I use CDROM_PACKET _COMMAND, the start address of the 12th track is 269850!? Why? I have have try several type discs, and all get the same problem: "a wrong start address of the last track". Is this a known bug or what? Or maybe I miss some details in the spec of ATAPI commands? The kernel running in my pc is 2.2.17. Should I patch or do something? Jack Ciao Nov 06

1 0

I would love to test packet but .....
by Eduardo M. A. M. Mendes 04 Nov '00

04 Nov '00

Hello First I would like to apologize for wasting the precious time of the experts. (A newbie can be quite annoying). However I really want to get my hands dirty as far as using udf-read-write on linux. Would it be a burden if I request some help from the experts? If not, please stay with me a little longer: a) What exactly are the files one should retrieve in order to get udf-write support on 2.4.0-test10? b) I couldn't follow the instructions for installing the packet software the last time I tried. May I humbly requested for some assistance on this matter? As result of it I can try to write a micro-howto on the subject (I hope I am not being too bold; if I am, please forgive me). c) Simple commands to test if the whole thing is working. That is all. Many many thanks for your time and patience. Yours Eduardo

1 0

cvs udf patch
by Jens Axboe 03 Nov '00

03 Nov '00

Hi, Another iteration of the timeout patch -- this one went to Ben privately, but more will probably want to see it before Ben commits it to the cvs tree (which he will no doubt have probably already done by now, since I took the time to write this e-mail (and considering his past promptness in checking stuff in for me)). -- * Jens Axboe <axboe(a)suse.de> * SuSE Labs

3 7

release: packet-0.0.2d
by Jens Axboe 01 Nov '00

01 Nov '00

Hello, I've just uploaded a new release of the packet writing patch, this time against the 2.4.0-test10 kernel. The bugs fixes that I've actually cared/remembered to write down are: - (scsi) use implicit segment recounting for all hba's - fix speed setting, was consistenly off on most drives - only print capacity when opening for write - fix off-by-two error in getting/setting write+read speed (affected reporting as well as actual speed used) - possible to enable write caching on drive - do ioctl marshalling on sparc64 from Ben Collins <bcollins(a)debian.org> - avoid unaligned access on flags, should have been unsigned long of course - fixed missed wakeup in kpacketd - b_dev error (two places) - fix buffer head b_count bugs - fix hole merge bug, where tail could be added twice - fsync and invalidate buffers on close - check hash table for buffers first before using our own - add read-ahead - fixed several list races - fix proc reporting for more than one device - change to O_CREAT for creating devices - added media_change hook - added free buffers config option - pkt_lock_tray fails on failed open (and oopses), remove it. unlock is done explicitly in pkt_remove dev anyway. - added proper elevator insertion (should probably be part of elevator.c) - moved kernel thread info to private device, spawn one for each writer - added separate buffer list for dirty packet buffers - fixed nasty data corruption bug - remember to account request even when we don't gather data for it - add ioctl to force wakeup of kernel thread (for debug) - fixed packet size setting bug on zero detected - changed a lot of the proc reporting to be more readable to "humans" - set full speed for read-only opens People wanting to give it a go, should also remember to update their UDF cvs tree (CDRW branch) and install new cdrwtool and pktsetup binaries. Interoperability with DirectCD has been tested, and that seems to work. Bugs / success stories (yeah right) should be repoted to the packet-writing list. *.kernel.org/pub/linux/kernel/people/axboe/packet/packet-0.0.2d.diff.bz2 -- * Jens Axboe <axboe(a)suse.de> * SuSE Labs

3 5

patch: stalls
by Jens Axboe 01 Nov '00

01 Nov '00

Hi, If you experience stalls when writing and the buffer cache is currently eating all of the available memory, this patch may help. Untested, I'm away from my machine atm. -- * Jens Axboe <axboe(a)suse.de> * SuSE Labs

1 0