Skip to content

uprobes: register/unregister bugfixes

Oleg Nesterov lkml 1 version In review

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 Latest7 patches 20 repliesIn review

Patches in v1 No applies recorded yet

#Subject and filesSizeRepliesApplied
1/7uprobes/x86: Only rep+nop can be emulated correctlyarch/x86/kernel/uprobes.c+2 -141 reply
2/7uprobes: Don't return success if alloc_uprobe() failskernel/events/uprobes.c+3 -11 reply
3/7uprobes: Do not delete uprobe if uprobe_unregister() failskernel/events/uprobes.c+6 -61 reply
4/7uprobes: Fix handle_swbp() vs unregister() + register() racekernel/events/uprobes.c+11 -06 replies
5/7uprobes: Introduce uprobe_copy_insn()include/linux/uprobes.h kernel/events/uprobes.c+41 -295 replies
6/7uprobes: Fix uprobe_copy_insn() race with itselfkernel/events/uprobes.c+8 -01 reply
7/7uprobes: Fix the racy uprobe->flags manipulationkernel/events/uprobes.c+14 -142 replies

Reviews trailers per patch

All 7 patches have a Reviewed-by or Acked-by.

PatchSrikar Dronamraju
1/7A-b
2/7A-b
3/7A-b
4/7A-b A-b A-b
5/7A-b A-b
6/7A-b
7/7A-b

Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.

Read the v1 thread

Cover letter [PATCH 0/7] uprobes: register/unregister bugfixes

Hello.

Misc bugfixes, mostly register/unregister related.

Note: 4/7 and especially 6/7 are the really ugly (but hopefully temporary) hacks. We already discussed this, we should inspect the original insn in uprobe_register(). But this is not easy, and we need something simple to fix (some of) the bugs right now.

Even 7/7 is ugly. Imho UPROBE_SKIP_SSTEP should die as well but this is another story.

Oleg.

 arch/x86/kernel/uprobes.c |   16 +------
 include/linux/uprobes.h   |   10 ----
 kernel/events/uprobes.c   |  107 +++++++++++++++++++++++++++++++-------------
 3 files changed, 77 insertions(+), 56 deletions(-)