Re: Re: [PATCH] md/raid1:combine code in fix_read_error andfix_sync_read_error.
From: majianpeng <hidden>
Date: 2012-04-10 01:17:41
Thanks,Neil.I did not notice the data buff is the same when write and read. ------------------ majianpeng 2012-04-10 ------------------------------------------------------------- 发件人：NeilBrown 发送日期：2012-04-10 07:47:15 收件人：majianpeng 抄送：linux-raid 主题：Re: [PATCH] md/raid1:combine code in fix_read_error andfix_sync_read_error. On Fri, 6 Apr 2012 15:15:38 +0800 "majianpeng" [off-list ref] wrote:
quoted
From d8d5044e65ed2bea15e135918699777c49b6f224 Mon Sep 17 00:00:00 2001From: majianpeng <redacted> Date: Fri, 6 Apr 2012 15:07:33 +0800 Subject: [PATCH] md/raid1:combine code in fix_read_error and fix_sync_read_error.
your explanation for the patch is, again, too short. I don't agree with this patch, I'd rather leave it as it is. I want to perform all the writes before any reads. If a read fails, I cannot be certain what the contents of the buffer are, so I don't want to risk writing it anywhere. Thanks, NeilBrown
quoted hunk ↗ jump to hunk
Signed-off-by: majianpeng <redacted> --- drivers/md/raid1.c | 36 ++++++++---------------------------- 1 files changed, 8 insertions(+), 28 deletions(-)diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c index d35e4c9..a9a4361 100644 --- a/drivers/md/raid1.c +++ b/drivers/md/raid1.c@@ -1603,7 +1603,6 @@ static int fix_sync_read_error(struct r1bio *r1_bio) int d = r1_bio->read_disk; int success = 0; struct md_rdev *rdev; - int start; if (s > (PAGE_SIZE>>9)) s = PAGE_SIZE >> 9;@@ -1661,7 +1660,6 @@ static int fix_sync_read_error(struct r1bio *r1_bio) continue; } - start = d; /* write it back and re-read */ while (d != r1_bio->read_disk) { if (d == 0)@@ -1675,21 +1673,12 @@ static int fix_sync_read_error(struct r1bio *r1_bio) WRITE) == 0) { r1_bio->bios[d]->bi_end_io = NULL; rdev_dec_pending(rdev, mddev); - } - } - d = start; - while (d != r1_bio->read_disk) { - if (d == 0) - d = conf->raid_disks * 2; - d--; - if (r1_bio->bios[d]->bi_end_io != end_sync_read) - continue; - rdev = conf->mirrors[d].rdev; - if (r1_sync_page_io(rdev, sect, s, - bio->bi_io_vec[idx].bv_page, - READ) != 0) + } else if (r1_sync_page_io(rdev, sect, s, + bio->bi_io_vec[idx].bv_page, READ)) atomic_add(s, &rdev->corrected_errors); + } + sectors -= s; sect += s; idx ++;@@ -1841,7 +1830,6 @@ static void fix_read_error(struct r1conf *conf, int read_disk, int s = sectors; int d = read_disk; int success = 0; - int start; struct md_rdev *rdev; if (s > (PAGE_SIZE>>9))@@ -1878,19 +1866,8 @@ static void fix_read_error(struct r1conf *conf, int read_disk, md_error(mddev, rdev); break; } + /* write it back and re-read */ - start = d; - while (d != read_disk) { - if (d==0) - d = conf->raid_disks * 2; - d--; - rdev = conf->mirrors[d].rdev; - if (rdev && - test_bit(In_sync, &rdev->flags)) - r1_sync_page_io(rdev, sect, s, - conf->tmppage, WRITE); - } - d = start; while (d != read_disk) { char b[BDEVNAME_SIZE]; if (d==0)@@ -1900,6 +1877,9 @@ static void fix_read_error(struct r1conf *conf, int read_disk, if (rdev && test_bit(In_sync, &rdev->flags)) { if (r1_sync_page_io(rdev, sect, s, + conf->tmppage, WRITE) == 0) + continue; + else if (r1_sync_page_io(rdev, sect, s, conf->tmppage, READ)) { atomic_add(s, &rdev->corrected_errors); printk(KERN_INFO