summaryrefslogtreecommitdiffstats
path: root/net/bridge/br_if.c
diff options
context:
space:
mode:
authorMike Snitzer <snitzer@kernel.org>2023-10-27 17:29:36 +0200
committerMike Snitzer <snitzer@kernel.org>2023-10-27 18:44:18 +0200
commit4a2fe2960891f1ccd7805d0973284fd44c2f12b4 (patch)
tree52572778a4f0036060f350e6ab25aff48e611c9b /net/bridge/br_if.c
parentdm: make __send_duplicate_bios return unsigned int (diff)
downloadlinux-4a2fe2960891f1ccd7805d0973284fd44c2f12b4.tar.xz
linux-4a2fe2960891f1ccd7805d0973284fd44c2f12b4.zip
dm: enhance alloc_multiple_bios() to be more versatile
alloc_multiple_bios() has the useful ability to try allocating bios with GFP_NOWAIT but will fallback to using GFP_NOIO. The callers service both empty flush bios and abnormal bios (e.g. discard). alloc_multiple_bios() enhancements offered in this commit: - don't require table_devices_lock if num_bios = 1 - allow caller to pass GFP_NOWAIT to do usual GFP_NOWAIT with GFP_NOIO fallback - allow caller to pass GFP_NOIO to _only_ allocate using GFP_NOIO Flush bios with data may be issued to DM with REQ_NOWAIT, as such it makes sense to attempt servicing them with GFP_NOWAIT allocations. But abnormal IO should never be issued using REQ_NOWAIT (if that changes in the future that's fine, but no sense supporting it now). While at it, rename __send_changing_extent_only() to __send_abnormal_io(). [Thanks to both Ming and Mikulas for help with translating known possible IO scenarios to requirements.] Suggested-by: Ming Lei <ming.lei@redhat.com> Suggested-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Diffstat (limited to 'net/bridge/br_if.c')
0 files changed, 0 insertions, 0 deletions