Currently zram runs compression and decompression in non-preemptible
sections, e.g.
zcomp_stream_get() // grabs CPU local lock
zcomp_compress()
or
zram_slot_lock() // grabs entry spin-lock
zcomp_stream_get() // grabs CPU local lock
zs_map_object() // grabs rwlock and CPU local lock
zcomp_decompress()
Potentially a little troublesome for a number of reasons.
For instance, this makes it impossible to use async compression
algorithms or/and H/W compression algorithms, which can wait for OP
completion or resource availability. This also restricts what
compression algorithms can do internally, for example, zstd can
allocate internal state memory for C/D dictionaries:
do_fsync()
do_writepages()
zram_bio_write()
zram_write_page() // become non-preemptible
zcomp_compress()
zstd_compress()
ZSTD_compress_usingCDict()
ZSTD_compressBegin_usingCDict_internal()
ZSTD_resetCCtx_usingCDict()
ZSTD_resetCCtx_internal()
zstd_custom_alloc() // memory allocation
Not to mention that the system can be configured to maximize
compression ratio at a cost of CPU/HW time (e.g. lz4hc or deflate
with very high compression level) so zram can stay in non-preemptible
section (even under spin-lock or/and rwlock) for an extended period
of time. Aside from compression algorithms, this also restricts what
zram can do. One particular example is zram_write_page() zsmalloc
handle allocation, which has an optimistic allocation (disallowing
direct reclaim) and a pessimistic fallback path, which then forces
zram to compress the page one more time.
This series changes zram to not directly impose atomicity restrictions
on compression algorithms (and on itself), which makes zram write()
fully preemptible; zram read(), sadly, is not always preemptible yet.
There are still indirect atomicity restrictions imposed by zsmalloc().
One notable example is object mapping API, which returns with:
a) local CPU lock held
b) zspage rwlock held
First, zsmalloc is converted to use sleepable RW-"lock" (it's atomic_t
in fact) for zspage migration protection. Second, a new handle mapping
is introduced which doesn't use per-CPU buffers (and hence no local CPU
lock), does fewer memcpy() calls, but requires users to provide a
pointer to temp buffer for object copy-in (when needed). Third, zram is
converted to the new zsmalloc mapping API and thus zram read() becomes
preemptible.
v4 -> v5:
- switched to preemptible per-CPU comp streams (Yosry)
- switched to preemptible bit-locks for zram entry locking (Andrew)
- added lockdep annotations to new zsmalloc/zram locks (Hillf, Yosry)
- perf measurements
- reworked re-compression loop (a bunch of minor fixes)
- fixed potential physical page leaks on writeback/recompression error
paths
- documented new locking rules
Sergey Senozhatsky (18):
zram: sleepable entry locking
zram: permit preemption with active compression stream
zram: remove crypto include
zram: remove max_comp_streams device attr
zram: remove two-staged handle allocation
zram: remove writestall zram_stats member
zram: limit max recompress prio to num_active_comps
zram: filter out recomp targets based on priority
zram: rework recompression loop
zsmalloc: factor out pool locking helpers
zsmalloc: factor out size-class locking helpers
zsmalloc: make zspage lock preemptible
zsmalloc: introduce new object mapping API
zram: switch to new zsmalloc object mapping API
zram: permit reclaim in zstd custom allocator
zram: do not leak page on recompress_store error path
zram: do not leak page on writeback_store error path
zram: add might_sleep to zcomp API
Documentation/ABI/testing/sysfs-block-zram | 8 -
Documentation/admin-guide/blockdev/zram.rst | 36 +-
drivers/block/zram/backend_zstd.c | 11 +-
drivers/block/zram/zcomp.c | 43 +-
drivers/block/zram/zcomp.h | 8 +-
drivers/block/zram/zram_drv.c | 286 +++++++------
drivers/block/zram/zram_drv.h | 22 +-
include/linux/zsmalloc.h | 8 +
mm/zsmalloc.c | 420 +++++++++++++++-----
9 files changed, 536 insertions(+), 306 deletions(-)
--
2.48.1.502.g6dc24dfdaf-goog
