BMC/Intel-BMC/linux.git/kernel/cgroup/cgroup-internal.h, branch dev-5.2

cgroup: make TRACE_CGROUP_PATH irq-safe

2019-04-19T18:26:49+00:00

To use the TRACE_CGROUP_PATH() macro with css_set_lock locked, let's make the macro irq-safe. It's necessary in order to trace cgroup freezer state transitions (frozen/not frozen), which are happening with css_set_lock locked. Signed-off-by: Roman Gushchin Signed-off-by: Tejun Heo

cgroup: implement __cgroup_task_count() helper

2019-04-19T18:26:48+00:00

The helper is identical to the existing cgroup_task_count() except it doesn't take the css_set_lock by itself, assuming that the caller does. Also, move cgroup_task_count() implementation into kernel/cgroup/cgroup.c, as there is nothing specific to cgroup v1. Signed-off-by: Roman Gushchin Signed-off-by: Tejun Heo Cc: kernel-team@fb.com

kernfs, sysfs, cgroup, intel_rdt: Support fs_context

2019-02-28T08:29:34+00:00

Make kernfs support superblock creation/mount/remount with fs_context. This requires that sysfs, cgroup and intel_rdt, which are built on kernfs, be made to support fs_context also. Notes: (1) A kernfs_fs_context struct is created to wrap fs_context and the kernfs mount parameters are moved in here (or are in fs_context). (2) kernfs_mount{,_ns}() are made into kernfs_get_tree(). The extra namespace tag parameter is passed in the context if desired (3) kernfs_free_fs_context() is provided as a destructor for the kernfs_fs_context struct, but for the moment it does nothing except get called in the right places. (4) sysfs doesn't wrap kernfs_fs_context since it has no parameters to pass, but possibly this should be done anyway in case someone wants to add a parameter in future. (5) A cgroup_fs_context struct is created to wrap kernfs_fs_context and the cgroup v1 and v2 mount parameters are all moved there. (6) cgroup1 parameter parsing error messages are now handled by invalf(), which allows userspace to collect them directly. (7) cgroup1 parameter cleanup is now done in the context destructor rather than in the mount/get_tree and remount functions. Weirdies: (*) cgroup_do_get_tree() calls cset_cgroup_from_root() with locks held, but then uses the resulting pointer after dropping the locks. I'm told this is okay and needs commenting. (*) The cgroup refcount web. This really needs documenting. (*) cgroup2 only has one root? Add a suggestion from Thomas Gleixner in which the RDT enablement code is placed into its own function. [folded a leak fix from Andrey Vagin] Signed-off-by: David Howells cc: Greg Kroah-Hartman cc: Tejun Heo cc: Li Zefan cc: Johannes Weiner cc: cgroups@vger.kernel.org cc: fenghua.yu@intel.com Signed-off-by: Al Viro

cgroup: store a reference to cgroup_ns into cgroup_fs_context

2019-02-28T08:29:34+00:00

... and trim cgroup_do_mount() arguments (renaming it to cgroup_do_get_tree()) Signed-off-by: Al Viro

cgroup_do_mount(): massage calling conventions

2019-02-28T08:29:33+00:00

pass it fs_context instead of fs_type/flags/root triple, have it return int instead of dentry and make it deal with setting fc->root. Signed-off-by: Al Viro

cgroup: stash cgroup_root reference into cgroup_fs_context

2019-02-28T08:29:32+00:00

Note that this reference is *NOT* contributing to refcount of cgroup_root in question and is valid only until cgroup_do_mount() returns. Signed-off-by: Al Viro

cgroup1: switch to option-by-option parsing

2019-02-28T08:29:31+00:00

[dhowells should be the author - it's carved out of his patch] Signed-off-by: Al Viro

cgroup: take options parsing into ->parse_monolithic()

2019-02-28T08:29:30+00:00

Store the results in cgroup_fs_context. There's a nasty twist caused by the enabling/disabling subsystems - we can't do the checks sensitive to that until cgroup_mutex gets grabbed. Frankly, these checks are complete bullshit (e.g. all,none combination is accepted if all subsystems are disabled; so's cpusets,none and all,cpusets when cpusets is disabled, etc.), but touching that would be a userland-visible behaviour change ;-/ So we do parsing in ->parse_monolithic() and have the consistency checks done in check_cgroupfs_options(), with the latter called (on already parsed options) from cgroup1_get_tree() and cgroup1_reconfigure(). Freeing the strdup'ed strings is done from fs_context destructor, which somewhat simplifies the life for cgroup1_{get_tree,reconfigure}(). Signed-off-by: Al Viro

cgroup: fold cgroup1_mount() into cgroup1_get_tree()

2019-02-28T08:29:30+00:00

Signed-off-by: Al Viro

cgroup: start switching to fs_context

2019-02-28T08:29:29+00:00

Unfortunately, cgroup is tangled into kernfs infrastructure. To avoid converting all kernfs-based filesystems at once, we need to untangle the remount part of things, instead of having it go through kernfs_sop_remount_fs(). Fortunately, it's not hard to do. This commit just gets cgroup/cgroup1 to use fs_context to deliver options on mount and remount paths. Parsing those is going to be done in the next commits; for now we do pretty much what legacy case does. Signed-off-by: Al Viro