iscsi error handler fixes

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



The following patches were made over Martin's 5.14 branch. They focus on
iscsi and scsi error handler bugs. This set combines the 2 patchsets I've
been posting to hopefully make it easier to backport just the in-kernel
fixes.

in-kernel conn failure handling fixes
-------------------------------------
V4:
- Fix case where we get multiple events for the same root issue, and userspace
and the kernel get out of sync due to multiple stops/cleanups

V3:
- Rebase over
https://urldefense.com/v3/__https://lore.kernel.org/linux-scsi/20210424220603.123703-1-michael.christie@xxxxxxxxxx/T/*t__;Iw!!GqivPVa7Brio!NdLMhOLAuEkbeyD-V8vbq1PDCQp51g1-HKalQU_YRA2dNB9qsEP4DAN8HvS0Dkwpflw-$
- Add fix for if we are doing offload boot and hit a conn failure at the
same time userspace is syncing up it's state and doing a relogin.
- Drop RFC as google has tested.

V2:
- Handle second part of #4 above and fix missing locking
- Include iscsi_tcp kernel sock shutdown patch

libiscsi and qedi TMF fixes
---------------------------
V5:
- Reduce number of error events sent for the same issue and handle
when libiscsi sends error events before we evern have a valid conn
binding.
- Drop tmf state check patch because we already were checking that
and qedi's session blocking was just plain not needed
- Drop cmd wait patch and replace with conn refcount patch
- Fix case where session reset could access a freed conn by grabbing
a ref to conn
- Have libiscsi handle iscsi_task refcounts during aborts for the
drivers

V4:
- Rm unused rtid variable in qedi
- Added a default case in TMF switch just in case people do not like
not having it.

V3:
- Fix u16 initialization and test.
- Fix bool return value use.
- Added patches for cases where EH is running then userspace terminates
the connection without removing the target first.
- Made patch that stops IO during ep_disconnect more driver friendly
by handling if the ep is bound or not.

V2:
- Dropped patch that reverted the in-kernel conn failure handling. I
posted a full patchset to fix that separately.
- Modfied the tmf vs cmd queueing patch. The RFC patch only supported
qedi and offload drivers. iscsi_tcp/cxgbgi do not need it.
- Added patch for another issue found where cmds can still be queued to the
driver while it does ep_disconnect.








[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Index of Archives]     [SCSI Target Devel]     [Linux SCSI Target Infrastructure]     [Kernel Newbies]     [IDE]     [Security]     [Git]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux ATA RAID]     [Linux IIO]     [Samba]     [Device Mapper]

  Powered by Linux