[PATCH v4 00/14] mm flags in printk, page_owner improvements for debugging

From: Vlastimil Babka
Date: Tue Jan 26 2016 - 07:51:47 EST

Changes since v3:
- Rebased on next-20160125
- Changed the %pg format string for flags to %pG due to clash with another
patch already merged merged
- Update GFP flags (patches 2, 3) for new changes and stuff I overlooked
the last time
- __GFP_X flags are now printed as __GFP_X instead of GFP_X as that was just
confusing and GFP_ATOMIC vs __GFP_ATOMIC already needed an exception

After v2 [1] of the page_owner series, I've moved mm-specific flags printing
into printk() and posted it on top of the series. But it makes more sense and
results in less code churn to do the printk() changes first. So this series
is two-part. Patches 1-6 are related to the flags handling in printk() and
tracepoints, and CC the printk, ftrace and perf maintainers. The rest is
the original page_owner series, CCing only mm people.

Adapted description of v2:

For page_owner, the main changes are
o Use static key to further reduce overhead when compiled in but not enabled.
o Improve output wrt. page and pageblock migratetypes
o Transfer the info on page migrations and track last migration reason.
o Dump the info as part of dump_page() to hopefully help debugging.

For the last point, Kirill requested a human readable printing of gfp_mask and
migratetype after v1. At that point it probably makes a lot of sense to do the
same for page alloc failure and OOM warnings. The flags have been undergoing
revisions recently, and we might be getting reports from various kernel
versions that differ. The ./scripts/gfp-translate tool needs to be pointed at
the corresponding sources to be accurate. The downside is potentially breaking
scripts that grep these warnings, but it's not a first change done there over
the years.

Other changes since v1:
o Change placement of page owner migration calls to cover missing cases (Hugh)
o Move dump_page_owner() call up from dump_page_badflags(), so the latter can
be used for adding debugging prints without page owner info (Kirill)

[1] https://lkml.org/lkml/2015/11/24/342

Vlastimil Babka (14):
tracepoints: move trace_print_flags definitions to tracepoint-defs.h
mm, tracing: make show_gfp_flags() up to date
tools, perf: make gfp_compact_table up to date
mm, tracing: unify mm flags handling in tracepoints and printk
mm, printk: introduce new format string for flags
mm, debug: replace dump_flags() with the new printk formats
mm, page_alloc: print symbolic gfp_flags on allocation failure
mm, oom: print symbolic gfp_flags in oom warning
mm, page_owner: print migratetype of page and pageblock, symbolic
mm, page_owner: convert page_owner_inited to static key
mm, page_owner: copy page owner info during migration
mm, page_owner: track and print last migrate reason
mm, page_owner: dump page owner info from dump_page()
mm, debug: move bad flags printing to bad_page()

Documentation/printk-formats.txt | 18 ++++
Documentation/vm/page_owner.txt | 9 +-
include/linux/gfp.h | 6 +-
include/linux/migrate.h | 6 +-
include/linux/mmdebug.h | 9 +-
include/linux/mmzone.h | 3 +
include/linux/page_ext.h | 1 +
include/linux/page_owner.h | 50 ++++++++---
include/linux/trace_events.h | 10 ---
include/linux/tracepoint-defs.h | 14 +++-
include/trace/events/btrfs.h | 2 +-
include/trace/events/compaction.h | 2 +-
include/trace/events/gfpflags.h | 43 ----------
include/trace/events/huge_memory.h | 2 -
include/trace/events/kmem.h | 2 +-
include/trace/events/mmflags.h | 165 +++++++++++++++++++++++++++++++++++++
include/trace/events/vmscan.h | 2 +-
lib/test_printf.c | 53 ++++++++++++
lib/vsprintf.c | 75 +++++++++++++++++
mm/debug.c | 165 +++++++++----------------------------
mm/internal.h | 6 ++
mm/migrate.c | 13 ++-
mm/oom_kill.c | 7 +-
mm/page_alloc.c | 29 +++++--
mm/page_owner.c | 100 +++++++++++++++++-----
mm/vmstat.c | 15 +---
tools/perf/builtin-kmem.c | 49 ++++++-----
27 files changed, 583 insertions(+), 273 deletions(-)
delete mode 100644 include/trace/events/gfpflags.h
create mode 100644 include/trace/events/mmflags.h