linux/fs/erofs/decompressor_deflate.c
Chunhai Guo d9281660ff erofs: relaxed temporary buffers allocation on readahead
Even with inplace decompression, sometimes very few temporary buffers
may be still needed for a single decompression shot (e.g. 16 pages for
64k sliding window or 4 pages for 16k sliding window).  In low-memory
scenarios, it would be better to try to allocate with GFP_NOWAIT on
readahead first.  That can help reduce the time spent on page allocation
under durative memory pressure.

Here are detailed performance numbers under multi-app launch benchmark
workload [1] on ARM64 Android devices (8-core CPU and 8GB of memory)
running a 5.15 LTS kernel with EROFS of 4k pclusters:

+----------------------------------------------+
|      LZ4       | vanilla | patched |  diff   |
|----------------+---------+---------+---------|
|  Average (ms)  |  3364   |  2684   | -20.21% | [64k sliding window]
|----------------+---------+---------+---------|
|  Average (ms)  |  2079   |  1610   | -22.56% | [16k sliding window]
+----------------------------------------------+

The total size of system images for 4k pclusters is almost unchanged:
(64k sliding window)  9,117,044 KB
(16k sliding window)  9,113,096 KB

Therefore, in addition to switch the sliding window from 64k to 16k,
after applying this patch, it can eventually save 52.14% (3364 -> 1610)
on average with no memory reservation.  That is particularly useful for
embedded devices with limited resources.

[1] https://lore.kernel.org/r/20240109074143.4138783-1-guochunhai@vivo.com

Suggested-by: Gao Xiang <xiang@kernel.org>
Signed-off-by: Chunhai Guo <guochunhai@vivo.com>
Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
Reviewed-by: Yue Hu <huyue2@coolpad.com>
Link: https://lore.kernel.org/r/20240126140142.201718-1-hsiangkao@linux.alibaba.com
2024-01-27 12:28:08 +08:00

255 lines
6.4 KiB
C

// SPDX-License-Identifier: GPL-2.0-or-later
#include <linux/zlib.h>
#include "compress.h"
struct z_erofs_deflate {
struct z_erofs_deflate *next;
struct z_stream_s z;
u8 bounce[PAGE_SIZE];
};
static DEFINE_SPINLOCK(z_erofs_deflate_lock);
static unsigned int z_erofs_deflate_nstrms, z_erofs_deflate_avail_strms;
static struct z_erofs_deflate *z_erofs_deflate_head;
static DECLARE_WAIT_QUEUE_HEAD(z_erofs_deflate_wq);
module_param_named(deflate_streams, z_erofs_deflate_nstrms, uint, 0444);
void z_erofs_deflate_exit(void)
{
/* there should be no running fs instance */
while (z_erofs_deflate_avail_strms) {
struct z_erofs_deflate *strm;
spin_lock(&z_erofs_deflate_lock);
strm = z_erofs_deflate_head;
if (!strm) {
spin_unlock(&z_erofs_deflate_lock);
continue;
}
z_erofs_deflate_head = NULL;
spin_unlock(&z_erofs_deflate_lock);
while (strm) {
struct z_erofs_deflate *n = strm->next;
vfree(strm->z.workspace);
kfree(strm);
--z_erofs_deflate_avail_strms;
strm = n;
}
}
}
int __init z_erofs_deflate_init(void)
{
/* by default, use # of possible CPUs instead */
if (!z_erofs_deflate_nstrms)
z_erofs_deflate_nstrms = num_possible_cpus();
for (; z_erofs_deflate_avail_strms < z_erofs_deflate_nstrms;
++z_erofs_deflate_avail_strms) {
struct z_erofs_deflate *strm;
strm = kzalloc(sizeof(*strm), GFP_KERNEL);
if (!strm)
goto out_failed;
/* XXX: in-kernel zlib cannot shrink windowbits currently */
strm->z.workspace = vmalloc(zlib_inflate_workspacesize());
if (!strm->z.workspace) {
kfree(strm);
goto out_failed;
}
spin_lock(&z_erofs_deflate_lock);
strm->next = z_erofs_deflate_head;
z_erofs_deflate_head = strm;
spin_unlock(&z_erofs_deflate_lock);
}
return 0;
out_failed:
erofs_err(NULL, "failed to allocate zlib workspace");
z_erofs_deflate_exit();
return -ENOMEM;
}
int z_erofs_load_deflate_config(struct super_block *sb,
struct erofs_super_block *dsb, void *data, int size)
{
struct z_erofs_deflate_cfgs *dfl = data;
if (!dfl || size < sizeof(struct z_erofs_deflate_cfgs)) {
erofs_err(sb, "invalid deflate cfgs, size=%u", size);
return -EINVAL;
}
if (dfl->windowbits > MAX_WBITS) {
erofs_err(sb, "unsupported windowbits %u", dfl->windowbits);
return -EOPNOTSUPP;
}
erofs_info(sb, "EXPERIMENTAL DEFLATE feature in use. Use at your own risk!");
return 0;
}
int z_erofs_deflate_decompress(struct z_erofs_decompress_req *rq,
struct page **pgpl)
{
const unsigned int nrpages_out =
PAGE_ALIGN(rq->pageofs_out + rq->outputsize) >> PAGE_SHIFT;
const unsigned int nrpages_in =
PAGE_ALIGN(rq->inputsize) >> PAGE_SHIFT;
struct super_block *sb = rq->sb;
unsigned int insz, outsz, pofs;
struct z_erofs_deflate *strm;
u8 *kin, *kout = NULL;
bool bounced = false;
int no = -1, ni = 0, j = 0, zerr, err;
/* 1. get the exact DEFLATE compressed size */
kin = kmap_local_page(*rq->in);
err = z_erofs_fixup_insize(rq, kin + rq->pageofs_in,
min_t(unsigned int, rq->inputsize,
sb->s_blocksize - rq->pageofs_in));
if (err) {
kunmap_local(kin);
return err;
}
/* 2. get an available DEFLATE context */
again:
spin_lock(&z_erofs_deflate_lock);
strm = z_erofs_deflate_head;
if (!strm) {
spin_unlock(&z_erofs_deflate_lock);
wait_event(z_erofs_deflate_wq, READ_ONCE(z_erofs_deflate_head));
goto again;
}
z_erofs_deflate_head = strm->next;
spin_unlock(&z_erofs_deflate_lock);
/* 3. multi-call decompress */
insz = rq->inputsize;
outsz = rq->outputsize;
zerr = zlib_inflateInit2(&strm->z, -MAX_WBITS);
if (zerr != Z_OK) {
err = -EIO;
goto failed_zinit;
}
pofs = rq->pageofs_out;
strm->z.avail_in = min_t(u32, insz, PAGE_SIZE - rq->pageofs_in);
insz -= strm->z.avail_in;
strm->z.next_in = kin + rq->pageofs_in;
strm->z.avail_out = 0;
while (1) {
if (!strm->z.avail_out) {
if (++no >= nrpages_out || !outsz) {
erofs_err(sb, "insufficient space for decompressed data");
err = -EFSCORRUPTED;
break;
}
if (kout)
kunmap_local(kout);
strm->z.avail_out = min_t(u32, outsz, PAGE_SIZE - pofs);
outsz -= strm->z.avail_out;
if (!rq->out[no]) {
rq->out[no] = erofs_allocpage(pgpl, rq->gfp);
if (!rq->out[no]) {
kout = NULL;
err = -ENOMEM;
break;
}
set_page_private(rq->out[no],
Z_EROFS_SHORTLIVED_PAGE);
}
kout = kmap_local_page(rq->out[no]);
strm->z.next_out = kout + pofs;
pofs = 0;
}
if (!strm->z.avail_in && insz) {
if (++ni >= nrpages_in) {
erofs_err(sb, "invalid compressed data");
err = -EFSCORRUPTED;
break;
}
if (kout) { /* unlike kmap(), take care of the orders */
j = strm->z.next_out - kout;
kunmap_local(kout);
}
kunmap_local(kin);
strm->z.avail_in = min_t(u32, insz, PAGE_SIZE);
insz -= strm->z.avail_in;
kin = kmap_local_page(rq->in[ni]);
strm->z.next_in = kin;
bounced = false;
if (kout) {
kout = kmap_local_page(rq->out[no]);
strm->z.next_out = kout + j;
}
}
/*
* Handle overlapping: Use bounced buffer if the compressed
* data is under processing; Or use short-lived pages from the
* on-stack pagepool where pages share among the same request
* and not _all_ inplace I/O pages are needed to be doubled.
*/
if (!bounced && rq->out[no] == rq->in[ni]) {
memcpy(strm->bounce, strm->z.next_in, strm->z.avail_in);
strm->z.next_in = strm->bounce;
bounced = true;
}
for (j = ni + 1; j < nrpages_in; ++j) {
struct page *tmppage;
if (rq->out[no] != rq->in[j])
continue;
DBG_BUGON(erofs_page_is_managed(EROFS_SB(sb),
rq->in[j]));
tmppage = erofs_allocpage(pgpl, rq->gfp);
if (!tmppage) {
err = -ENOMEM;
goto failed;
}
set_page_private(tmppage, Z_EROFS_SHORTLIVED_PAGE);
copy_highpage(tmppage, rq->in[j]);
rq->in[j] = tmppage;
}
zerr = zlib_inflate(&strm->z, Z_SYNC_FLUSH);
if (zerr != Z_OK || !(outsz + strm->z.avail_out)) {
if (zerr == Z_OK && rq->partial_decoding)
break;
if (zerr == Z_STREAM_END && !outsz)
break;
erofs_err(sb, "failed to decompress %d in[%u] out[%u]",
zerr, rq->inputsize, rq->outputsize);
err = -EFSCORRUPTED;
break;
}
}
failed:
if (zlib_inflateEnd(&strm->z) != Z_OK && !err)
err = -EIO;
if (kout)
kunmap_local(kout);
failed_zinit:
kunmap_local(kin);
/* 4. push back DEFLATE stream context to the global list */
spin_lock(&z_erofs_deflate_lock);
strm->next = z_erofs_deflate_head;
z_erofs_deflate_head = strm;
spin_unlock(&z_erofs_deflate_lock);
wake_up(&z_erofs_deflate_wq);
return err;
}