kernel/drivers/of/overlay.c, branch linux-4.16.y

of: Use SPDX license tag for DT files

2018-01-08T14:22:45Z

Convert remaining DT files to use SPDX-License-Identifier tags. Cc: Benjamin Herrenschmidt Cc: Guennadi Liakhovetski Cc: Paul Mackerras Cc: Pantelis Antoniou Reviewed-by: Frank Rowand Reviewed-by: Philippe Ombredanne Signed-off-by: Rob Herring

of: overlay: Make node skipping in init_overlay_changeset() clearer

2017-12-08T15:32:18Z

Make it more clear that nodes without "__overlay__" subnodes are skipped, by reverting the logic and using continue. This also reduces indentation level. Signed-off-by: Geert Uytterhoeven Signed-off-by: Rob Herring

of: overlay: Fix out-of-bounds write in init_overlay_changeset()

2017-12-08T15:32:10Z

If an overlay has no "__symbols__" node, but it has nodes without "__overlay__" subnodes at the end (e.g. a "__fixups__" node), after filling in all fragments for nodes with "__overlay__" subnodes, "fragment = &fragments[cnt]" will point beyond the end of the allocated array. Hence writing to "fragment->overlay" will overwrite unallocated memory, which may lead to a crash later. Fix this by deferring both the assignment to "fragment" and the offending write afterwards until we know for sure the node has an "__overlay__" subnode, and thus a valid entry in "fragments[]". Fixes: 61b4de4e0b384f4a ("of: overlay: minor restructuring") Signed-off-by: Geert Uytterhoeven Signed-off-by: Rob Herring

of: overlay: Fix (un)locking in of_overlay_apply()

2017-12-06T22:08:30Z

The special overlay mutex is taken first, hence it should be released last in the error path. of_resolve_phandles() must be called with of_mutex held. Without it, a node and new phandle could be added via of_attach_node(), making the max phandle wrong. free_overlay_changeset() must be called with of_mutex held, if any non-trivial cleanup is to be done. Hence move "mutex_lock(&of_mutex)" up, as suggested by Frank, and merge the two tail statements of the success and error paths, now they became identical. Note that while the two mutexes are adjacent, we still need both: __of_changeset_apply_notify(), which is called by __of_changeset_apply() unlocks of_mutex, then does notifications then locks of_mutex. So the mutex get released in the middle of of_overlay_apply() Fixes: f948d6d8b792bb90 ("of: overlay: avoid race condition between applying multiple overlays") Signed-off-by: Geert Uytterhoeven Reviewed-by: Frank Rowand Signed-off-by: Rob Herring

of: overlay: Fix memory leak in of_overlay_apply() error path

2017-12-06T22:04:36Z

If of_resolve_phandles() fails, free_overlay_changeset() is called in the error path. However, that function returns early if the list hasn't been initialized yet, before freeing the object. Explicitly calling kfree() instead would solve that issue. However, that complicates matter, by having to consider which of two different methods to use to dispose of the same object. Hence make free_overlay_changeset() consider initialization state of the different parts of the object, making it always safe to call (once!) to dispose of a (partially) initialized overlay_changeset: - Only destroy the changeset if the list was initialized, - Make init_overlay_changeset() store the ID in ovcs->id on success, to avoid calling idr_remove() with an error value or an already released ID. Reported-by: Colin King Fixes: f948d6d8b792bb90 ("of: overlay: avoid race condition between applying multiple overlays") Signed-off-by: Geert Uytterhoeven Reviewed-by: Frank Rowand Signed-off-by: Rob Herring

of: overlay: Remove else after goto

2017-12-06T20:56:33Z

If an "if" branch is terminated by a "goto", there's no need to have an "else" statement and an indented block of code. Remove the "else" statement to simplify the code flow for the casual reviewer. Signed-off-by: Geert Uytterhoeven Signed-off-by: Rob Herring

of: Spelling s/changset/changeset/

2017-12-06T20:56:32Z

Signed-off-by: Geert Uytterhoeven Signed-off-by: Rob Herring

of: overlay: make pr_err() string unique

2017-10-19T22:15:09Z

The same error string occurs in drivers/of/resolver.c. Change the error here to more precisely describe this case, and avoid the possible confusion of looking in the wrong source location to understand the cause of an error. Signed-off-by: Frank Rowand Signed-off-by: Rob Herring

of: overlay: pr_err from return NOTIFY_OK to overlay apply/remove

2017-10-19T22:15:09Z

A device tree overlay notifier can return NOTIFY_OK, NOTIFY_STOP, or an embedded errno. overlay_notify() incorrectly reports an error for NOTIFY_OK. Reported-by: atull@kernel.org Signed-off-by: Frank Rowand Signed-off-by: Rob Herring

of: overlay: remove unneeded check for NULL kbasename()

2017-10-18T01:47:56Z

kbasename() will not return NULL if passed a valid string. If the parameter passed to kbasename() in this case is already NULL then the devicetree has been corrupted. Signed-off-by: Frank Rowand Signed-off-by: Rob Herring