| Message ID | 20220519140021.6905-1-hare@suse.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | blk-cgroup: provide stubs for blkcg_get_fc_appid() | expand |
On Thu, May 19, 2022 at 04:00:21PM +0200, Hannes Reinecke wrote: > Provide stubs for blkcg_set_fc_appid() and blkcg_get_fc_appid() to allow > for compilation with cgroups disabled. > > Fixes: db05628435aa ("blk-cgroup: move blkcg_{get,set}_fc_appid out of line") > Signed-off-by: Hannes Reinecke <hare@suse.de> No, it does not fix that commit, which is perfectly fine. It fixes the recently added second caller of blkcg_get_fc_appid, and James has just resent a new version of that which fixes this properly.
On 5/19/22 07:08, Christoph Hellwig wrote: > On Thu, May 19, 2022 at 04:00:21PM +0200, Hannes Reinecke wrote: >> Provide stubs for blkcg_set_fc_appid() and blkcg_get_fc_appid() to allow >> for compilation with cgroups disabled. >> >> Fixes: db05628435aa ("blk-cgroup: move blkcg_{get,set}_fc_appid out of line") >> Signed-off-by: Hannes Reinecke <hare@suse.de> > > No, it does not fix that commit, which is perfectly fine. It fixes > the recently added second caller of blkcg_get_fc_appid, and James > has just resent a new version of that which fixes this properly. Really? blk-cgroup.h provides the function declaration blkcg_get_fc_appid() unconditionally, but the implementation for blkcg_get_fc_appid() depends on CONFIG_CGROUP. Neither of which is changed by James patchset. And besides, the first version is already merged. Am I missing something? Cheers, Hannes
On 5/19/22 8:00 AM, Hannes Reinecke wrote: > Provide stubs for blkcg_set_fc_appid() and blkcg_get_fc_appid() to allow > for compilation with cgroups disabled. > > Fixes: db05628435aa ("blk-cgroup: move blkcg_{get,set}_fc_appid out of line") This isn't a valid sha.
On Thu, May 19, 2022 at 04:20:40PM +0200, Hannes Reinecke wrote: >> >> No, it does not fix that commit, which is perfectly fine. It fixes >> the recently added second caller of blkcg_get_fc_appid, and James >> has just resent a new version of that which fixes this properly. > > Really? blk-cgroup.h provides the function declaration > blkcg_get_fc_appid() unconditionally, but the implementation > for blkcg_get_fc_appid() depends on CONFIG_CGROUP. Take a look at the IS_ENABLED macro that we have for a few years now. > Neither of which is changed by James patchset. > > And besides, the first version is already merged. > > Am I missing something? https://lore.kernel.org/linux-scsi/20220519123110.17361-1-jsmart2021@gmail.com/T/#t
diff --git a/include/linux/blk-cgroup.h b/include/linux/blk-cgroup.h index 9f40dbc65f82..4756f4d2b8e7 100644 --- a/include/linux/blk-cgroup.h +++ b/include/linux/blk-cgroup.h @@ -33,6 +33,9 @@ void blkcg_unpin_online(struct cgroup_subsys_state *blkcg_css); struct list_head *blkcg_get_cgwb_list(struct cgroup_subsys_state *css); struct cgroup_subsys_state *bio_blkcg_css(struct bio *bio); +int blkcg_set_fc_appid(char *app_id, u64 cgrp_id, size_t app_id_len); +char *blkcg_get_fc_appid(struct bio *bio); + #else /* CONFIG_BLK_CGROUP */ #define blkcg_root_css ((struct cgroup_subsys_state *)ERR_PTR(-EINVAL)) @@ -44,9 +47,15 @@ static inline struct cgroup_subsys_state *bio_blkcg_css(struct bio *bio) { return NULL; } +static inline int blkcg_set_fc_appid(char *app_id, u64 cgrp_id, + size_t app_id_len) +{ + return -EINVAL; +} +static inline char *blkcg_get_fc_appid(struct bio *bio) +{ + return NULL; +} #endif /* CONFIG_BLK_CGROUP */ -int blkcg_set_fc_appid(char *app_id, u64 cgrp_id, size_t app_id_len); -char *blkcg_get_fc_appid(struct bio *bio); - #endif /* _BLK_CGROUP_H */
Provide stubs for blkcg_set_fc_appid() and blkcg_get_fc_appid() to allow for compilation with cgroups disabled. Fixes: db05628435aa ("blk-cgroup: move blkcg_{get,set}_fc_appid out of line") Signed-off-by: Hannes Reinecke <hare@suse.de> --- include/linux/blk-cgroup.h | 15 ++++++++++++--- 1 file changed, 12 insertions(+), 3 deletions(-)