On 01/12/2019 08:45, Coly Li wrote:
On 2019/12/1 6:34 上午, Eric Wheeler wrote:
Hi Coly,
We use 4.19.y and there have been many performance and stability changes
since then. I'm considering backporting the 5.4 version into 4.19 and
wondered:
Are there any changes in bcache between 4.19 and 5.4 that depend on new
features elsewhere in the kernel, or should I basically be able to copy
the tree from 5.4 to 4.19 and fix minor compilation issues?
Can you think of any issues that would arise from such a backport?
Hi Eric,
It should be OK to backport bcache patches from 5.4 to 4.19. I did
similar thing for SUSE Enterprise kernel which were based on Linux 4.12
code base, and the changes was tiny.
If you encounter problem during the backport, you may post the .rej file
and maybe I can help.
Hi Eric, Coly,
I have been applying the list below to kernel.org 4.19.x latest stable
and using it on several production machines without any issues for
months apart from the newer ones of course. Apart from the odd context
adjust I don't think any needed any modifying. My list used to be much
bigger but a lot has been backported already in 4.19.x stable so my list
has gone down quite a lot :-)
Sorry I don't use git but if I run in my patches directory as it is
currently the list is:
egrep '^Subject' *.patch
00a.patch:Subject: bcache: account size of buckets used in uuid write to
00b.patch:Subject: bcache: fix typo in code comments of
closure_return_with_destructor()
00c.patch:Subject: bcache: recal cached_dev_sectors on detach
01.patch:Subject: bcache: remove unused bch_passthrough_cache
02.patch:Subject: bcache: remove useless parameter of bch_debug_init()
03.patch:Subject: bcache: use MAX_CACHES_PER_SET instead of magic number
8 in
05.patch:Subject: bcache: do not check if debug dentry is ERR or NULL
explicitly on
06.patch:Subject: bcache: do not mark writeback_running too early
07.patch:Subject: bcache: cannot set writeback_running via sysfs if no
writeback
08.patch:Subject: bcache: introduce force_wake_up_gc()
09.patch:Subject: bcache: not use hard coded memset size in
11.patch:Subject: bcache: fix input integer overflow of congested threshold
13.patch:Subject: bcache: add sysfs_strtoul_bool() for setting bit-field
variables
14.patch:Subject: bcache: use sysfs_strtoul_bool() to set bit-field
variables
15.patch:Subject: bcache: fix input overflow to writeback_delay
17.patch:Subject: bcache: fix input overflow to journal_delay_ms
18.patch:Subject: bcache: fix input overflow to cache set io_error_limit
20.patch:Subject: bcache: fix crashes stopping bcache device before read
miss done
22.patch:Subject: bcache: fix inaccurate result of unused buckets
24.patch:Subject: bcache: add error check for calling register_bdev()
26.patch:Subject: bcache: improve bcache_reboot()
28.patch:Subject: bcache: fix wrong usage use-after-freed on keylist in
out_nocoalesce
30.patch:Subject: [PATCH 07/12] bcache: fix deadlock in bcache_allocator
