September 2000 - Packet Writing - openSUSE Mailing Lists

linux-udf patch
by Mark Robinson 30 Sep '00

30 Sep '00

Hey, I am having trouble applying the udf patch from packet-0.0.2d-pre1 to the linux-udf source tree I retrived via cvs. Actually, when I cd to the newly created udf directory, and call patch < ../packet-0.0.2d/udf-cdrw-1509.diff, I get the following error message: *** can't find file to patch at input line 8 Perhaps you should have used the -p or --strip option? The text leading up to this was: -------------------------- |Index: linux-2.4/balloc.c |=================================================================== |RCS file: /cvsroot/linux-udf/udf/linux-2.4/balloc.c,v |retrieving revision 1.12.2.3 |diff -u -r1.12.2.3 balloc.c |--- linux-2.4/balloc.c 2000/06/21 18:30:19 1.12.2.3 |+++ linux-2.4/balloc.c 2000/09/15 01:02:33 -------------------------- File to patch: *** If I type the name in (in this case, linux-2.4/balloc.c), the patch applies. I get to a point where it wants to patch files in the src subdirectory that do not exist in the udf source tree. What gives? Regards, Mark Robinson ____________NetZero Free Internet Access and Email_________ Download Now http://www.netzero.net/download/index.html Request a CDROM 1-800-333-3633 ___________________________________________________________

1 0

Re: [Re: [packet-writing] Re: packet-writing with Redhat?]
by Glenn M 24 Sep '00

24 Sep '00

Hi Adam, Well I had some sucess but also some real problems. Adam <adam(a)cfar.umd.edu> wrote: > mount -t udf -o rw,user /dev/pktcdvd0 /mnt/cdrom > mount: block device /dev/pktcdvd0 is write-protected, mounting read-only mabye b/c you compiled udf w/out "Write Support"? In order to get you need to enable 'experimental options' and then enabled 'UDF Write (Dangerous)" Correct, I did not have this set. After I recompiled I could mount it and did write to the CD. > 1) Why aren't udf and pktcdvd0 modules being auto loaded even though > they appear in my modules.dep file? b/c modules.dep just lists intermodule dependencies. In order to have something call the module kernel/knod or whatver it is nowdays needs to know what to call for example if I do 'ifconfig eth0' then the "whatever" will check out /etc/modules.conf what eth0 is defined as Yeah, forgot about this. Exactly what do you have? alias pktcdvd0 alias /dev/scdo udf OK, now the big problems. 1) After I recompiled I could not make a new initrd image. It just hung. I killed the bash process. 2) After umount, I could not eject the CD 3) I could not shutdown. I had to use the reset switch - ugh Glenn ----- ____________________________________________________________________ Get free email and a permanent address at http://www.netaddress.com/?N=1

2 1

Re: current state of packet writing ??
by Glenn 24 Sep '00

24 Sep '00

Glenn wrote: Hello Jens and all, OK that did the trick for UDF. I had already compiled the test7 kernel with your patch and just made the udf module and did the modprobe thing. No problems yet. I was also able to format a CDRW just fine. My question/problem is in getting the new device to be recognized. I created it: mknod /dev/pktcdvd0 b 97 0 and verified that it existed after this problem: # ./pktsetup /dev/pktcdvd0 /dev/scd0 packet open: No such device [root@h000094b6613f /packages/udf/tools] I also get this when I tried to mount it, recognizing that this is probably foolish since it didn't set it up, but don't understand the file system not being supported since the module "udf" is installed and active. # mount -t udf /dev/pktcdvd0 /mnt/cdr mount: fs type udf not supported by kernel [root@h000094b6613f /dev] Appreciate any help that can be offered. Thanks, Glenn ----- > > Jens Axboe wrote: > > > > On Mon, Sep 18 2000, Jens Axboe wrote: > > > You have an out-of-date UDF tree. For use with packet writing, current > > > CDRW branch of the UDF tree is recommended. > > > > Probably should have mentioned that the UDF page to go to is > > > > http://sourceforge.net/projects/linux-udf/ > > > > -- > > * Jens Axboe <axboe(a)suse.de> > > * SuSE Labs

4 7

Re: current state of packet writing ??
by Jens Axboe 18 Sep '00

18 Sep '00

On Sat, Sep 16 2000, Glenn wrote: > Hi, > > I'm trying to use this package but am unable to create the udftools > package with the test7 kernel. I have Redhat beta 7 installed, which is > called "pinstripe". The latest version of the tools that I could find > relies on earlier versions of the kernel. > > # make udftools > > --- snip > > Kernel configuration options: > Symmetric multiprocessing support is disabled. > Max physical memory in MB is > Advanced Power Management (APM) support is disabled. > Networking support is enabled. > Module version checking is enabled. > > The kernel source tree is version 2.4.0-test7. > The current kernel build date is Sat Sep 16 10:06:23 EDT 2000. > This package requires a 2.2/2.3 series kernel. > make: *** [.prereq.ok] Error 1 > [root@h000094b6613f /packages/udf-0.9.2.1] You have an out-of-date UDF tree. For use with packet writing, current CDRW branch of the UDF tree is recommended. -- * Jens Axboe <axboe(a)suse.de> * SuSE Labs

1 1

0.0.2d-pre1, further notes
by Jens Axboe 15 Sep '00

15 Sep '00

Hi, Just before leaving, I remembered something else that might be worth to try. When init every packet writer queue to hold 32 free requests, to put less pressure on the device it may be worthwhile to try 2 instead (one for reads, one for writes). Maybe this will affect the corruption issue, even if not it will be easier to troubleshoot that way. Look for blk_init_queue_nr(q, pkt_request, 32); and change to blk_init_queue_nr(q, pkt_request, 2); -- * Jens Axboe <axboe(a)suse.de> * SuSE Labs

1 0

packet-0.0.2d-pre1
by Jens Axboe 15 Sep '00

15 Sep '00

Hi, Adam encouraged me to release a 0.0.2d even if I was gone, as someone else might be able to fix it (or at least play with it). So here it is. In order to aid nailing this problem I've written it down from my view - so that either I discover the bug while explaining it, or at least give some one an idea of what to look for. The bugs manifests itself by causing corruption of some of the data written to disc. It _appears_ that this corruption only happens when the queue is unplugged due to memory pressue (buffer cache is filled, and something needs to be written out). The corruption looks like data ending up in the wrong places. Copying several big .bz2 files to disc and might show some of the blocks from file1 in file2 and so on. I removed the buffer hole-merge support, even though I couldn't see any bugs in that. The corruption was still there, so that does not appear to be the culprit. This pre-release does not have the hole merge support back in, for this reason. I even went as far as md5 summing data when I had the entire request but before scheduling it for another queue, and then comparing each individual buffer when it was completed (in the end_io handler). Nothing showed up there. That wasn't it, so at least data is consistent from when we let it go and when the drive has written it out (unless the drive is garbling data, not very likely though...). I tried this because I thought that someone may be using the buffer even though we have it locked, but that doesn't not appear to be what is happening. Other changes since 0.0.2c include: * - (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 * - 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 support * - added free buffers config option So there you have it. If you feel like bug hunting or just playing around a bit, it's there. Note that you need to compile and use a new pktsetup as well! Use of UDF CDRW branch is recommended, I've put a put for it within the pre release. *.kernel.org/pub/linux/kernel/people/axboe/packet/test/ -- * Jens Axboe <axboe(a)suse.de> * SuSE Labs

1 0

next release
by Jens Axboe 14 Sep '00

14 Sep '00

Hi folks, I'm still struggling with a corruption issue and that's delaying 0.0.2d, I'm afraid. In addition, I'll be gone until monday so don't expect anything before the start of next week. -- * Jens Axboe <axboe(a)suse.de> * SuSE Labs

1 0

strange problem
by Alexander V. Trotsay 14 Sep '00

14 Sep '00

Hi all I'm try to make disk with cdrwtool All Ok, I have 538Mb free space blank disk Than I mount it and without any writes umount When I try mount disk again I got strange information pktcdvd: 754KB available on disc and I can't mount disk there some debug from dmesg UDF-fs DEBUG (lowlevel.c, 65): udf_get_last_session: CDROMMULTISESSION not supported: rc=-22 UDF-fs DEBUG (super.c, 1316): udf_read_super: Multi-session=0 attempt to access beyond end of device 61:00: rw=0, want=2147483590, limit=755 attempt to access beyond end of device 61:00: rw=0, want=2147483478, limit=755 attempt to access beyond end of device 61:00: rw=0, want=2147483586, limit=755 attempt to access beyond end of device 61:00: rw=0, want=2147483474, limit=755 attempt to access beyond end of device 61:00: rw=0, want=0, limit=755 attempt to access beyond end of device 61:00: rw=0, want=2147483464, limit=755 attempt to access beyond end of device 61:00: rw=0, want=2147483352, limit=755 attempt to access beyond end of device 61:00: rw=0, want=2147483460, limit=755 attempt to access beyond end of device 61:00: rw=0, want=2147483348, limit=755 UDF-fs DEBUG (super.c, 1325): udf_read_super: Lastblock=0 UDF-fs DEBUG (super.c, 389): udf_vrs: Starting at sector 16 (2048 byte sectors) UDF-fs DEBUG (super.c, 739): udf_load_pvoldesc: recording time 968662169/236618, 2000/09/11 11:49 (10b4) UDF-fs DEBUG (super.c, 856): udf_load_logicalvol: Partition (0) type 2 UDF-fs DEBUG (super.c, 912): udf_load_logicalvol: FileSet found in LogicalVolDesc at block=32, partition=0 __________________________________________________________________ Info&Contacts -- RIPE: AT252-RIPE, InterNic: AT2963, ICQ: 18035130 My skill in net: http://www.brainbench.com/transcript.jsp?pid=232045 Trouble of the day: Operators killed by year 2000 bug bite.

2 4

strange problem 2
by Alexander V. Trotsay 11 Sep '00

11 Sep '00

Hi again Than I try remove disk and insert it again Disk mounted (strange) After this I try to write some file to this disk As for me file wrote while buffer space availiable, than mc (or cp, I try both) got seg fault signal and died. I try to wrote many small files, may be trouble with this. And I have some kernel output like this Device b00 not ready. I/O error: dev 0b:00, sector 12532 Device b00 not ready. I/O error: dev 0b:00, sector 12536 Device b00 not ready. I/O error: dev 0b:00, sector 12540 assert failed sectors == rq->nr_sectors,pkt_recheck_segments at 142 kernel BUG at pktcdvd.c:142! invalid operand: 0000 CPU: 0 EIP: 0010:[<c018df32>] EFLAGS: 00010286 eax: 0000001d ebx: 00000080 ecx: 00000001 edx: 00000000 esi: 0000001b edi: c292cd20 ebp: c292cd20 esp: c7f75f14 ds: 0018 es: 0018 ss: 0018 Process kpacketd (pid: 6, stackpage=c7f75000) Stack: c0251d05 c0251f0e 0000008e c4885e60 c4885e60 00003100 c018e833 c292cd20 c6e3cc40 0000002c 00003100 00003080 00000001 c1252400 c1252400 00003246 c292cd20 c7f77000 c292cc60 c7f77058 c018e911 c7f77000 c292cd20 c292cd20 Call Trace: [<c0251d05>] [<c0251f0e>] [<c018e833>] [<c018e911>] [<c018ea00>] [<c018eb9f>] [<c018eaa0>] [<c0108c3b>] Code: 0f 0b 83 c4 0c 5b 5e 5f c3 90 8d 74 26 00 55 8b 0d 6c af 2e and device now blocked, I can't do nothing and may I need make unclean reboot __________________________________________________________________ Info&Contacts -- RIPE: AT252-RIPE, InterNic: AT2963, ICQ: 18035130 My skill in net: http://www.brainbench.com/transcript.jsp?pid=232045 Trouble of the day: The Usenet news is out of date

1 0

udf code?
by Ravi K. Swamy 10 Sep '00

10 Sep '00

Which UDF code should be used with the packet writing code? Should I enable CONFIG_UDF_FS and CONFIG_UDF_RW in the kernel config or use the udf.o module from the 0.9.2.1 sourceforge CVS tree? The install instructions are a little unclear since they say: - Select packet support in the block device section and UDF support in the file system section. - Make UDF (make lib tools udf.o install) ^^^^^ Thanks, Ravi

3 3