These patches contain the following linux kernel bug fixes:
Description:
cio: I/O can not be stopped after vary off.
Symptom:
Memory corruption (oops) after qeth chpid has been logically
varied off while network traffic is running.
Problem:
qeth tries to do clears on its read and qdio devices before
removing its card structures. However, this does not work since
the devices are in state WAIT4IO which blocks I/O requests from
the device driver while waiting for running I/O to terminate.
qeth does not check the return code and throws away its card
structures. Since the queues are still up, incoming traffic
may overwrite memory areas before they are killed by cio after
the grace period is over.
Solution:
Allow device drivers to issue ccw_device_halt() and
ccw_device_clear() during the grace period.
Problem-ID:
13698
Description:
kernel: swap size is limited to 4GB.
Symptom:
A swap disk is used only up to 4GB on a 64 bit linux
system although free reports a bigger swap space.
Problem:
The mk_swap_pte primitive only allows 20 bits in the
swap offset. This is the correct limit for 31 bit
but not for 64 bit.
Solution:
Allow 53 bits in the swap offset for 64 bit.
Problem-ID:
13883
Note:
applicable for 64-bit Linux, only
Description:
qeth: dhcpcd does not work.
Symptom:
dhcpcd cannot communicate with DHCP server.
Problem:
Incorrect packet headers when fake_ll is active.
Solution:
Establish qeth_fake_header function and
activate it in the hard_header field of the net_device structure
if fake_ll is active.
Omit specific fake_ll processing for outbound raw packets
(protocol neither IP nor IPV6).
Problem-ID:
7141
Description:
qeth: qethconf ipa list failed after adding a vipa or parp
address.
Symptom:
Unable to handle kernel pointer dereference at virtual kernel
address 00000000.
Problem:
Memory corruption because of an invalid variable assignment.
Solution:
Fixed variable assignment.
Problem-ID:
13601
Description:
tape: Tar or DB/2 backup hangs.
Symptom:
Writing to tape hangs when tape library sends notification about
statistics log change.
Problem:
Unsolicited interrupt after starting I/O can cause a deferred
condition code. In that case we have to restart the last request.
Solution:
Check for status pending and start function not executed and
restart I/O in that case.
Problem-ID:
12854
Description:
zfcp: Error in handler of FSF requests 'Send FCP Command'.
Symptom:
If 'Send FCP Command' request comes back with FSF status
FSF_ADAPTER_STATUS_AVAILABLE and unknown FSF status qualifier
then this request is not marked in ZFCP module as bad request.
Problem:
Send FCP Command handler marks request as bad for FSF status
FSF_ADAPTER_STATUS_AVAILABLE only for known FSF status qualifiers.
Solution:
Mark request as bad in any case if FSF status is
FSF_ADAPTER_STATUS_AVAILABLE.
Problem-ID:
13816
Everybody should apply this patch.
To create the complete linux kernel sources, the following
patches need to be applied in sequence:
