Kernel patches
2003 bug fix patches
2003 kerntypes patches
2003 On-demand timer patches
2003 toolchain patches
2003 utilitity patches
- gdb-5.3-s390
- modutils-2.4.7-s390
- strace-4.4-1-s390
- strace-4.4-1-s390-01
- strace-4.4-1-s390-02
- lkcdutils-4.1 Reduce dump size (savedump) patch
- lkcdutils-4.1 lkcd dump format version 8 patch
2004 bug fix patches
2004 On-demand timer patches
2005 bug fix patches
2006 bug fix patches
2006 On-demand timer patches
2003-07-25 kernel 2.4.21 bug fix patch 03 ("June 2003")
If you download any software from this web site please be aware of the Warranty Disclaimer and Limitation of Liabilities.
linux-2.4.21-s390-03-june2003.tar.gz / MD5 ... accumulated patch, recommended (2003-07-25)
linux-2.4.21-s390-03-june2003-patches.tar.gz / MD5 ... per-problem-patches, recommended (2003-07-25)
These patches contain the following linux kernel bug fixes:
- Description:
- xpram parameter parsing.
- Symptom:
- The xpram driver doesn't accept hexadecimal sizes on the xpram_parts parameter with a single digit after the 0x.
- Problem:
- The check in xpram_int_format that determines if a number is in hexadecimal is broken.
- Solution:
- Fix the check for hexadecimal numbers.
- Problem-ID:
- 3284
- Description:
- Use z990 path group algorithm also on 64 bit
- Note:
- applicable for 64-bit Linux, only
- Description:
- spinlock problem with unsupported ioctls in qeth
- Symptom:
- deadlock in ioctl of unsupported ioctl of qeth
- Problem:
- spinlock was not released
- Solution:
- relese spinlock before leaving function
- Problem-ID:
- 3287
- Description:
- unloading qeth module results in a kernel oops while osasnmpd is running
- Symptom:
- 1. osasnmpd is running
2. shut down all network interfaces related to qeth module
3. rmmod qeth causes a kernel oops - Problem:
- osasnmpd has an open ioctl on /proc/qeth without incrementing qeth module usage counter
- Solution:
- increment the qeth module usage counter when opening the ioctl and decrement it when leaving the function and so returning to osasnmpd
- Problem-ID:
- 3317
- Description:
- sysinfo doesn't return all data on 31 bit emu.
- Symptom:
- mem_unit info missing.
- Problem:
- Some data members were not copied in emu layer.
- Solution:
- Copy of data members added.
- Problem-ID:
- n/a
- Note:
- applicable for 64-bit Linux, only
- Description:
- cio: ctcs not detected; varying of network chpids broken.
- Symptom:
- 1. ctcs are not detected if the other side of the connection is
active.
2. When varying a networking chpid off and then on again via /proc/chpids, the networking interface does not recover. - Problem:
- 1. SNID is not supported on ctcs, the command code means
something else. This leads to ctcs being detected as boxed.
2. The channel device layer cannot handle DEVSTAT_NOT_ACC. - Solution:
- 1. Don't do a SNID as first action during device recognition.
If a SID causes timeout, consider the device boxed. Only do
SNID/SPID if the device driver requests it during
request_irq.
2. Only use DEVSTAT_NOT_ACC for device drivers which requested SNID/SPID. Use DEVSTAT_DEVICE_GONE for other device drivers. - Problem-ID:
- 3395
- Description:
- cio: link incident handling.
- Symptom:
- After receiving a link incident, Linux hangs.
- Problem:
- The event information stored after a link incident contains a reporting chpid; this may or may not be the chpid with the link incident.
- Solution:
- Check the link incident record received for the chpid with the link incident.
- Problem-ID:
- 3398
- Description:
- <ccwcache>: Incorrect cache index determination
- Symptom:
- random kernel oops
- Problem:
- The incorrect comparison caused the wrong cache line to be selected for maximum size requests.
- Solution:
- fixed the comparison
- Problem-ID:
- 3450
- Description:
- siginfo_t structure on user stack incorrect.
- Symptom:
- Unwinding over a rt signal frame crashes user programs.
- Problem:
- The kernel definition of the siginfo_t structure for s390x is 136 bytes long instead of the correct length of 128. The kernel will create a stack frame for rt signal that the unwind code can't process.
- Solution:
- Reduce the size of the _pad array.
- Problem-ID:
- 3490
- Note:
- Description:
- cannot rmmod lcs module when TR-devices are used
- Symptom:
- rmmod lcs does not work, when ipv6 and TR were used
- Problem:
- IPv6 gets some stray packets and looses track of dev->refcnt
- Solution:
- make lcs not give v6 packets to the stack
- Problem-ID:
- 3491
- Description:
- HW checksum fix
- Symptom:
- HW checksumming did not work
- Problem:
- problem in bit usage of QDIO header
- Solution:
- interpret QDIO header correctly
- Problem-ID:
- n/a
- Description:
- reduce stack usage
- Symptom:
- symlink loops on NFS share causes oops
- Problem:
- stack overflow
- Solution:
- reduce stack usage by inlining some functions
- Problem-ID:
- 3344
- Description:
- added qeth options to Configure.help
- Description:
- added z90crypt options to Configure.help
Everybody should apply this patch.
To create the complete linux kernel sources, the following patches need to be applied in sequence:
linux-2.4.21.tar.gz (see www.kernel.org)
+ linux-2.4.21-s390-june2003.diff (IBM)
+ linux-2.4.21-s390-01-june2003.diff (IBM)
+ linux-2.4.21-s390-02-june2003.diff (IBM)
+ linux-2.4.21-s390-03-june2003.diff (IBM)
Note: If On-demand timer is required, apply as last patch:
+ linux-2.4.21-s390-timer-june2003.diff (IBM)