From: Andrew Morton <akpm@linux-foundation.org>
To: mm-commits@vger.kernel.org,minchan@kernel.org,gaoxu2@honor.com,bgeffon@google.com,axboe@kernel.dk,senozhatsky@chromium.org,akpm@linux-foundation.org
Subject: + zram-remove-chained-recompression.patch added to mm-unstable branch
Date: Wed, 11 Mar 2026 11:12:55 -0700 [thread overview]
Message-ID: <20260311181255.A12FAC4CEF7@smtp.kernel.org> (raw)
The patch titled
Subject: zram: remove chained recompression
has been added to the -mm mm-unstable branch. Its filename is
zram-remove-chained-recompression.patch
This patch will shortly appear at
https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patches/zram-remove-chained-recompression.patch
This patch will later appear in the mm-unstable branch at
git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
Before you just go and hit "reply", please:
a) Consider who else should be cc'ed
b) Prefer to cc a suitable mailing list as well
c) Ideally: find the original patch on the mailing list and do a
reply-to-all to that, adding suitable additional cc's
*** Remember to use Documentation/process/submit-checklist.rst when testing your code ***
The -mm tree is included into linux-next via various
branches at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
and is updated there most days
------------------------------------------------------
From: Sergey Senozhatsky <senozhatsky@chromium.org>
Subject: zram: remove chained recompression
Date: Wed, 11 Mar 2026 17:42:48 +0900
Chained recompression has unpredictable behavior and is not useful in
practice.
First, systems usually configure just one alternative recompression
algorithm, which has slower compression/decompression but better
compression ratio. A single alternative algorithm doesn't need chaining.
Second, even with multiple recompression algorithms, chained recompression
is suboptimal. If a lower priority algorithm succeeds, the page is never
attempted with a higher priority algorithm, leading to worse memory
savings. If a lower priority algorithm fails, the page is still attempted
with a higher priority algorithm, wasting resources on the failed lower
priority attempt.
In either case, the system would be better off targeting a specific
priority directly.
Chained recompression also significantly complicates the code. Remove it.
Link: https://lkml.kernel.org/r/20260311084312.1766036-6-senozhatsky@chromium.org
Signed-off-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Cc: Brian Geffon <bgeffon@google.com>
Cc: gao xu <gaoxu2@honor.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---
Documentation/admin-guide/blockdev/zram.rst | 9 -
drivers/block/zram/zram_drv.c | 84 +++++-------------
2 files changed, 24 insertions(+), 69 deletions(-)
--- a/Documentation/admin-guide/blockdev/zram.rst~zram-remove-chained-recompression
+++ a/Documentation/admin-guide/blockdev/zram.rst
@@ -533,15 +533,6 @@ unexpected results when the same algorit
priorities (e.g. different parameters). `priority` is the only way to
guarantee that the expected algorithm will be used.
-During re-compression for every page, that matches re-compression criteria,
-ZRAM iterates the list of registered alternative compression algorithms in
-order of their priorities. ZRAM stops either when re-compression was
-successful (re-compressed object is smaller in size than the original one)
-and matches re-compression criteria (e.g. size threshold) or when there are
-no secondary algorithms left to try. If none of the secondary algorithms can
-successfully re-compressed the page such a page is marked as incompressible,
-so ZRAM will not attempt to re-compress it in the future.
-
memory tracking
===============
--- a/drivers/block/zram/zram_drv.c~zram-remove-chained-recompression
+++ a/drivers/block/zram/zram_drv.c
@@ -2347,7 +2347,7 @@ static bool highest_priority_algorithm(s
return true;
}
-static int scan_slots_for_recompress(struct zram *zram, u32 mode, u32 prio_max,
+static int scan_slots_for_recompress(struct zram *zram, u32 mode, u32 prio,
struct zram_pp_ctl *ctl)
{
unsigned long nr_pages = zram->disksize >> PAGE_SHIFT;
@@ -2373,8 +2373,8 @@ static int scan_slots_for_recompress(str
test_slot_flag(zram, index, ZRAM_INCOMPRESSIBLE))
goto next;
- /* Already compressed with same of higher priority */
- if (get_slot_comp_priority(zram, index) + 1 >= prio_max)
+ /* Already compressed with same or higher priority */
+ if (get_slot_comp_priority(zram, index) >= prio)
goto next;
ok = place_pp_slot(zram, ctl, index);
@@ -2395,8 +2395,7 @@ next:
* Corresponding ZRAM slot should be locked.
*/
static int recompress_slot(struct zram *zram, u32 index, struct page *page,
- u64 *num_recomp_pages, u32 threshold, u32 prio,
- u32 prio_max)
+ u64 *num_recomp_pages, u32 threshold, u32 prio)
{
struct zcomp_strm *zstrm = NULL;
unsigned long handle_old;
@@ -2408,6 +2407,9 @@ static int recompress_slot(struct zram *
void *src;
int ret = 0;
+ if (!zram->comps[prio])
+ return -EINVAL;
+
handle_old = get_slot_handle(zram, index);
if (!handle_old)
return -EINVAL;
@@ -2430,51 +2432,10 @@ static int recompress_slot(struct zram *
*/
clear_slot_flag(zram, index, ZRAM_IDLE);
- class_index_old = zs_lookup_class_index(zram->mem_pool, comp_len_old);
-
- prio = max(prio, get_slot_comp_priority(zram, index) + 1);
- /*
- * Recompression slots scan should not select slots that are
- * already compressed with a higher priority algorithm, but
- * just in case
- */
- if (prio >= prio_max)
- return 0;
-
- /*
- * Iterate the secondary comp algorithms list (in order of priority)
- * and try to recompress the page.
- */
- for (; prio < prio_max; prio++) {
- if (!zram->comps[prio])
- continue;
-
- zstrm = zcomp_stream_get(zram->comps[prio]);
- src = kmap_local_page(page);
- ret = zcomp_compress(zram->comps[prio], zstrm,
- src, &comp_len_new);
- kunmap_local(src);
-
- if (ret) {
- zcomp_stream_put(zstrm);
- zstrm = NULL;
- break;
- }
-
- class_index_new = zs_lookup_class_index(zram->mem_pool,
- comp_len_new);
-
- /* Continue until we make progress */
- if (class_index_new >= class_index_old ||
- (threshold && comp_len_new >= threshold)) {
- zcomp_stream_put(zstrm);
- zstrm = NULL;
- continue;
- }
-
- /* Recompression was successful so break out */
- break;
- }
+ zstrm = zcomp_stream_get(zram->comps[prio]);
+ src = kmap_local_page(page);
+ ret = zcomp_compress(zram->comps[prio], zstrm, src, &comp_len_new);
+ kunmap_local(src);
/*
* Decrement the limit (if set) on pages we can recompress, even
@@ -2485,11 +2446,18 @@ static int recompress_slot(struct zram *
if (*num_recomp_pages)
*num_recomp_pages -= 1;
- /* Compression error */
- if (ret)
+ if (ret) {
+ zcomp_stream_put(zstrm);
return ret;
+ }
+
+ class_index_old = zs_lookup_class_index(zram->mem_pool, comp_len_old);
+ class_index_new = zs_lookup_class_index(zram->mem_pool, comp_len_new);
+
+ if (class_index_new >= class_index_old ||
+ (threshold && comp_len_new >= threshold)) {
+ zcomp_stream_put(zstrm);
- if (!zstrm) {
/*
* Secondary algorithms failed to re-compress the page
* in a way that would save memory.
@@ -2539,11 +2507,11 @@ static ssize_t recompress_store(struct d
struct device_attribute *attr,
const char *buf, size_t len)
{
- u32 prio = ZRAM_SECONDARY_COMP, prio_max = ZRAM_MAX_COMPS;
struct zram *zram = dev_to_zram(dev);
char *args, *param, *val, *algo = NULL;
u64 num_recomp_pages = ULLONG_MAX;
struct zram_pp_ctl *ctl = NULL;
+ u32 prio = ZRAM_SECONDARY_COMP;
struct zram_pp_slot *pps;
u32 mode = 0, threshold = 0;
struct page *page = NULL;
@@ -2597,8 +2565,6 @@ static ssize_t recompress_store(struct d
ret = kstrtouint(val, 10, &prio);
if (ret)
return ret;
-
- prio_max = min(prio + 1, ZRAM_MAX_COMPS);
continue;
}
}
@@ -2618,7 +2584,6 @@ static ssize_t recompress_store(struct d
continue;
if (!strcmp(zram->comp_algs[prio], algo)) {
- prio_max = min(prio + 1, ZRAM_MAX_COMPS);
found = true;
break;
}
@@ -2647,7 +2612,7 @@ static ssize_t recompress_store(struct d
goto out;
}
- scan_slots_for_recompress(zram, mode, prio_max, ctl);
+ scan_slots_for_recompress(zram, mode, prio, ctl);
ret = len;
while ((pps = select_pp_slot(ctl))) {
@@ -2661,8 +2626,7 @@ static ssize_t recompress_store(struct d
goto next;
err = recompress_slot(zram, pps->index, page,
- &num_recomp_pages, threshold,
- prio, prio_max);
+ &num_recomp_pages, threshold, prio);
next:
slot_unlock(zram, pps->index);
release_pp_slot(zram, pps);
_
Patches currently in -mm which might be from senozhatsky@chromium.org are
zram-do-not-permit-params-change-after-init.patch
zram-do-not-autocorrect-bad-recompression-parameters.patch
zram-drop-num_active_comps.patch
zram-update-recompression-documentation.patch
zram-remove-chained-recompression.patch
zram-unify-and-harden-algo-priority-params-handling.patch
reply other threads:[~2026-03-11 18:12 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20260311181255.A12FAC4CEF7@smtp.kernel.org \
--to=akpm@linux-foundation.org \
--cc=axboe@kernel.dk \
--cc=bgeffon@google.com \
--cc=gaoxu2@honor.com \
--cc=minchan@kernel.org \
--cc=mm-commits@vger.kernel.org \
--cc=senozhatsky@chromium.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.