| Message ID | 20250414111141.735081-1-chao@kernel.org (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | [f2fs-dev] resize.f2fs: fix to always change metadata for expand resize | expand |
Hm. Would this be what @uplinkr might have encountered? On Mon, Apr 14, 2025 at 4:13 AM Chao Yu via Linux-f2fs-devel <linux-f2fs-devel@lists.sourceforge.net> wrote: > > w/ below testcase, resize will generate a corrupted image which > contains inconsistent metadata: > > touch img > truncate -s $((512*1024*1024*1024)) img > mkfs.f2fs -f img $((256*1024*1024)) > resize.f2fs -s img -t $((1024*1024*1024)) > mount img /mnt/f2fs > > [ 31.725200] F2FS-fs (loop0): Wrong bitmap size: sit: 192, sit_blk_cnt:4762 > [ 31.728441] F2FS-fs (loop0): Failed to get valid F2FS checkpoint > > The root cause is safe mode (via -s option) is not compatible > w/ expand resize, due to in safe mode, we will keep all parameters > related to NAT, SIT, SSA area, e.g. sit_bitmap_size, however, we > will update segment_count_main according new partition size, result > in there is no enough sit_bitmap and SIT blocks to address the > entire block address of new partition. > > Adding a check condition to avoid expanding partition in safe > mode, and change manual accordingly. > > Signed-off-by: Chao Yu <chao@kernel.org> > --- > fsck/resize.c | 12 ++++++++---- > man/resize.f2fs.8 | 2 +- > 2 files changed, 9 insertions(+), 5 deletions(-) > > diff --git a/fsck/resize.c b/fsck/resize.c > index 1ab7d75..58914ec 100644 > --- a/fsck/resize.c > +++ b/fsck/resize.c > @@ -756,18 +756,22 @@ int f2fs_resize(struct f2fs_sb_info *sbi) > > /* may different sector size */ > if ((c.target_sectors * c.sector_size >> > - get_sb(log_blocksize)) < get_sb(block_count)) > + get_sb(log_blocksize)) < get_sb(block_count)) { > if (!c.safe_resize) { > ASSERT_MSG("Nothing to resize, now only supports resizing with safe resize flag\n"); > return -1; > } else { > return f2fs_resize_shrink(sbi); > } > - else if (((c.target_sectors * c.sector_size >> > + } else if (((c.target_sectors * c.sector_size >> > get_sb(log_blocksize)) > get_sb(block_count)) || > - c.ignore_error) > + c.ignore_error) { > + if (c.safe_resize) { > + ASSERT_MSG("Expanding resize doesn't support safe resize flag"); > + return -1; > + } > return f2fs_resize_grow(sbi); > - else { > + } else { > MSG(0, "Nothing to resize.\n"); > return 0; > } > diff --git a/man/resize.f2fs.8 b/man/resize.f2fs.8 > index bdda4fd..5b6daf5 100644 > --- a/man/resize.f2fs.8 > +++ b/man/resize.f2fs.8 > @@ -69,7 +69,7 @@ Skip caution dialogue and resize partition directly. > Specify support write hint. > .TP > .BI \-s > -Enable safe resize. > +Enable safe resize, it can only be used w/ shrink resize. > .TP > .BI \-V > Print the version number and exit. > -- > 2.49.0 > > > > _______________________________________________ > Linux-f2fs-devel mailing list > Linux-f2fs-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
On 4/16/25 03:28, Juhyung Park wrote: > Hm. > > Would this be what @uplinkr might have encountered? Maybe, :) @uplinkr, previously, if you used '-s' parameter while expand-resizing, that could be the reason it corrupted your partition. Thanks, > > On Mon, Apr 14, 2025 at 4:13 AM Chao Yu via Linux-f2fs-devel > <linux-f2fs-devel@lists.sourceforge.net> wrote: >> >> w/ below testcase, resize will generate a corrupted image which >> contains inconsistent metadata: >> >> touch img >> truncate -s $((512*1024*1024*1024)) img >> mkfs.f2fs -f img $((256*1024*1024)) >> resize.f2fs -s img -t $((1024*1024*1024)) >> mount img /mnt/f2fs >> >> [ 31.725200] F2FS-fs (loop0): Wrong bitmap size: sit: 192, sit_blk_cnt:4762 >> [ 31.728441] F2FS-fs (loop0): Failed to get valid F2FS checkpoint >> >> The root cause is safe mode (via -s option) is not compatible >> w/ expand resize, due to in safe mode, we will keep all parameters >> related to NAT, SIT, SSA area, e.g. sit_bitmap_size, however, we >> will update segment_count_main according new partition size, result >> in there is no enough sit_bitmap and SIT blocks to address the >> entire block address of new partition. >> >> Adding a check condition to avoid expanding partition in safe >> mode, and change manual accordingly. >> >> Signed-off-by: Chao Yu <chao@kernel.org> >> --- >> fsck/resize.c | 12 ++++++++---- >> man/resize.f2fs.8 | 2 +- >> 2 files changed, 9 insertions(+), 5 deletions(-) >> >> diff --git a/fsck/resize.c b/fsck/resize.c >> index 1ab7d75..58914ec 100644 >> --- a/fsck/resize.c >> +++ b/fsck/resize.c >> @@ -756,18 +756,22 @@ int f2fs_resize(struct f2fs_sb_info *sbi) >> >> /* may different sector size */ >> if ((c.target_sectors * c.sector_size >> >> - get_sb(log_blocksize)) < get_sb(block_count)) >> + get_sb(log_blocksize)) < get_sb(block_count)) { >> if (!c.safe_resize) { >> ASSERT_MSG("Nothing to resize, now only supports resizing with safe resize flag\n"); >> return -1; >> } else { >> return f2fs_resize_shrink(sbi); >> } >> - else if (((c.target_sectors * c.sector_size >> >> + } else if (((c.target_sectors * c.sector_size >> >> get_sb(log_blocksize)) > get_sb(block_count)) || >> - c.ignore_error) >> + c.ignore_error) { >> + if (c.safe_resize) { >> + ASSERT_MSG("Expanding resize doesn't support safe resize flag"); >> + return -1; >> + } >> return f2fs_resize_grow(sbi); >> - else { >> + } else { >> MSG(0, "Nothing to resize.\n"); >> return 0; >> } >> diff --git a/man/resize.f2fs.8 b/man/resize.f2fs.8 >> index bdda4fd..5b6daf5 100644 >> --- a/man/resize.f2fs.8 >> +++ b/man/resize.f2fs.8 >> @@ -69,7 +69,7 @@ Skip caution dialogue and resize partition directly. >> Specify support write hint. >> .TP >> .BI \-s >> -Enable safe resize. >> +Enable safe resize, it can only be used w/ shrink resize. >> .TP >> .BI \-V >> Print the version number and exit. >> -- >> 2.49.0 >> >> >> >> _______________________________________________ >> Linux-f2fs-devel mailing list >> Linux-f2fs-devel@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
Just checked gparted's source code: https://github.com/GNOME/gparted/blob/GPARTED_1_7_0/src/f2fs.cc#L135 Seems unlikely. :/ On Tue, Apr 15, 2025 at 8:34 PM Chao Yu <chao@kernel.org> wrote: > > On 4/16/25 03:28, Juhyung Park wrote: > > Hm. > > > > Would this be what @uplinkr might have encountered? > > Maybe, :) > > @uplinkr, previously, if you used '-s' parameter while expand-resizing, > that could be the reason it corrupted your partition. > > Thanks, > > > > > On Mon, Apr 14, 2025 at 4:13 AM Chao Yu via Linux-f2fs-devel > > <linux-f2fs-devel@lists.sourceforge.net> wrote: > >> > >> w/ below testcase, resize will generate a corrupted image which > >> contains inconsistent metadata: > >> > >> touch img > >> truncate -s $((512*1024*1024*1024)) img > >> mkfs.f2fs -f img $((256*1024*1024)) > >> resize.f2fs -s img -t $((1024*1024*1024)) > >> mount img /mnt/f2fs > >> > >> [ 31.725200] F2FS-fs (loop0): Wrong bitmap size: sit: 192, sit_blk_cnt:4762 > >> [ 31.728441] F2FS-fs (loop0): Failed to get valid F2FS checkpoint > >> > >> The root cause is safe mode (via -s option) is not compatible > >> w/ expand resize, due to in safe mode, we will keep all parameters > >> related to NAT, SIT, SSA area, e.g. sit_bitmap_size, however, we > >> will update segment_count_main according new partition size, result > >> in there is no enough sit_bitmap and SIT blocks to address the > >> entire block address of new partition. > >> > >> Adding a check condition to avoid expanding partition in safe > >> mode, and change manual accordingly. > >> > >> Signed-off-by: Chao Yu <chao@kernel.org> > >> --- > >> fsck/resize.c | 12 ++++++++---- > >> man/resize.f2fs.8 | 2 +- > >> 2 files changed, 9 insertions(+), 5 deletions(-) > >> > >> diff --git a/fsck/resize.c b/fsck/resize.c > >> index 1ab7d75..58914ec 100644 > >> --- a/fsck/resize.c > >> +++ b/fsck/resize.c > >> @@ -756,18 +756,22 @@ int f2fs_resize(struct f2fs_sb_info *sbi) > >> > >> /* may different sector size */ > >> if ((c.target_sectors * c.sector_size >> > >> - get_sb(log_blocksize)) < get_sb(block_count)) > >> + get_sb(log_blocksize)) < get_sb(block_count)) { > >> if (!c.safe_resize) { > >> ASSERT_MSG("Nothing to resize, now only supports resizing with safe resize flag\n"); > >> return -1; > >> } else { > >> return f2fs_resize_shrink(sbi); > >> } > >> - else if (((c.target_sectors * c.sector_size >> > >> + } else if (((c.target_sectors * c.sector_size >> > >> get_sb(log_blocksize)) > get_sb(block_count)) || > >> - c.ignore_error) > >> + c.ignore_error) { > >> + if (c.safe_resize) { > >> + ASSERT_MSG("Expanding resize doesn't support safe resize flag"); > >> + return -1; > >> + } > >> return f2fs_resize_grow(sbi); > >> - else { > >> + } else { > >> MSG(0, "Nothing to resize.\n"); > >> return 0; > >> } > >> diff --git a/man/resize.f2fs.8 b/man/resize.f2fs.8 > >> index bdda4fd..5b6daf5 100644 > >> --- a/man/resize.f2fs.8 > >> +++ b/man/resize.f2fs.8 > >> @@ -69,7 +69,7 @@ Skip caution dialogue and resize partition directly. > >> Specify support write hint. > >> .TP > >> .BI \-s > >> -Enable safe resize. > >> +Enable safe resize, it can only be used w/ shrink resize. > >> .TP > >> .BI \-V > >> Print the version number and exit. > >> -- > >> 2.49.0 > >> > >> > >> > >> _______________________________________________ > >> Linux-f2fs-devel mailing list > >> Linux-f2fs-devel@lists.sourceforge.net > >> https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel >
On 4/16/25 11:40, Juhyung Park wrote: > Just checked gparted's source code: > https://github.com/GNOME/gparted/blob/GPARTED_1_7_0/src/f2fs.cc#L135 > > Seems unlikely. :/ Alright, seems still we need another testcase for reproducing. :( Thanks, > > On Tue, Apr 15, 2025 at 8:34 PM Chao Yu <chao@kernel.org> wrote: >> >> On 4/16/25 03:28, Juhyung Park wrote: >>> Hm. >>> >>> Would this be what @uplinkr might have encountered? >> >> Maybe, :) >> >> @uplinkr, previously, if you used '-s' parameter while expand-resizing, >> that could be the reason it corrupted your partition. >> >> Thanks, >> >>> >>> On Mon, Apr 14, 2025 at 4:13 AM Chao Yu via Linux-f2fs-devel >>> <linux-f2fs-devel@lists.sourceforge.net> wrote: >>>> >>>> w/ below testcase, resize will generate a corrupted image which >>>> contains inconsistent metadata: >>>> >>>> touch img >>>> truncate -s $((512*1024*1024*1024)) img >>>> mkfs.f2fs -f img $((256*1024*1024)) >>>> resize.f2fs -s img -t $((1024*1024*1024)) >>>> mount img /mnt/f2fs >>>> >>>> [ 31.725200] F2FS-fs (loop0): Wrong bitmap size: sit: 192, sit_blk_cnt:4762 >>>> [ 31.728441] F2FS-fs (loop0): Failed to get valid F2FS checkpoint >>>> >>>> The root cause is safe mode (via -s option) is not compatible >>>> w/ expand resize, due to in safe mode, we will keep all parameters >>>> related to NAT, SIT, SSA area, e.g. sit_bitmap_size, however, we >>>> will update segment_count_main according new partition size, result >>>> in there is no enough sit_bitmap and SIT blocks to address the >>>> entire block address of new partition. >>>> >>>> Adding a check condition to avoid expanding partition in safe >>>> mode, and change manual accordingly. >>>> >>>> Signed-off-by: Chao Yu <chao@kernel.org> >>>> --- >>>> fsck/resize.c | 12 ++++++++---- >>>> man/resize.f2fs.8 | 2 +- >>>> 2 files changed, 9 insertions(+), 5 deletions(-) >>>> >>>> diff --git a/fsck/resize.c b/fsck/resize.c >>>> index 1ab7d75..58914ec 100644 >>>> --- a/fsck/resize.c >>>> +++ b/fsck/resize.c >>>> @@ -756,18 +756,22 @@ int f2fs_resize(struct f2fs_sb_info *sbi) >>>> >>>> /* may different sector size */ >>>> if ((c.target_sectors * c.sector_size >> >>>> - get_sb(log_blocksize)) < get_sb(block_count)) >>>> + get_sb(log_blocksize)) < get_sb(block_count)) { >>>> if (!c.safe_resize) { >>>> ASSERT_MSG("Nothing to resize, now only supports resizing with safe resize flag\n"); >>>> return -1; >>>> } else { >>>> return f2fs_resize_shrink(sbi); >>>> } >>>> - else if (((c.target_sectors * c.sector_size >> >>>> + } else if (((c.target_sectors * c.sector_size >> >>>> get_sb(log_blocksize)) > get_sb(block_count)) || >>>> - c.ignore_error) >>>> + c.ignore_error) { >>>> + if (c.safe_resize) { >>>> + ASSERT_MSG("Expanding resize doesn't support safe resize flag"); >>>> + return -1; >>>> + } >>>> return f2fs_resize_grow(sbi); >>>> - else { >>>> + } else { >>>> MSG(0, "Nothing to resize.\n"); >>>> return 0; >>>> } >>>> diff --git a/man/resize.f2fs.8 b/man/resize.f2fs.8 >>>> index bdda4fd..5b6daf5 100644 >>>> --- a/man/resize.f2fs.8 >>>> +++ b/man/resize.f2fs.8 >>>> @@ -69,7 +69,7 @@ Skip caution dialogue and resize partition directly. >>>> Specify support write hint. >>>> .TP >>>> .BI \-s >>>> -Enable safe resize. >>>> +Enable safe resize, it can only be used w/ shrink resize. >>>> .TP >>>> .BI \-V >>>> Print the version number and exit. >>>> -- >>>> 2.49.0 >>>> >>>> >>>> >>>> _______________________________________________ >>>> Linux-f2fs-devel mailing list >>>> Linux-f2fs-devel@lists.sourceforge.net >>>> https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel >>
diff --git a/fsck/resize.c b/fsck/resize.c index 1ab7d75..58914ec 100644 --- a/fsck/resize.c +++ b/fsck/resize.c @@ -756,18 +756,22 @@ int f2fs_resize(struct f2fs_sb_info *sbi) /* may different sector size */ if ((c.target_sectors * c.sector_size >> - get_sb(log_blocksize)) < get_sb(block_count)) + get_sb(log_blocksize)) < get_sb(block_count)) { if (!c.safe_resize) { ASSERT_MSG("Nothing to resize, now only supports resizing with safe resize flag\n"); return -1; } else { return f2fs_resize_shrink(sbi); } - else if (((c.target_sectors * c.sector_size >> + } else if (((c.target_sectors * c.sector_size >> get_sb(log_blocksize)) > get_sb(block_count)) || - c.ignore_error) + c.ignore_error) { + if (c.safe_resize) { + ASSERT_MSG("Expanding resize doesn't support safe resize flag"); + return -1; + } return f2fs_resize_grow(sbi); - else { + } else { MSG(0, "Nothing to resize.\n"); return 0; } diff --git a/man/resize.f2fs.8 b/man/resize.f2fs.8 index bdda4fd..5b6daf5 100644 --- a/man/resize.f2fs.8 +++ b/man/resize.f2fs.8 @@ -69,7 +69,7 @@ Skip caution dialogue and resize partition directly. Specify support write hint. .TP .BI \-s -Enable safe resize. +Enable safe resize, it can only be used w/ shrink resize. .TP .BI \-V Print the version number and exit.
w/ below testcase, resize will generate a corrupted image which contains inconsistent metadata: touch img truncate -s $((512*1024*1024*1024)) img mkfs.f2fs -f img $((256*1024*1024)) resize.f2fs -s img -t $((1024*1024*1024)) mount img /mnt/f2fs [ 31.725200] F2FS-fs (loop0): Wrong bitmap size: sit: 192, sit_blk_cnt:4762 [ 31.728441] F2FS-fs (loop0): Failed to get valid F2FS checkpoint The root cause is safe mode (via -s option) is not compatible w/ expand resize, due to in safe mode, we will keep all parameters related to NAT, SIT, SSA area, e.g. sit_bitmap_size, however, we will update segment_count_main according new partition size, result in there is no enough sit_bitmap and SIT blocks to address the entire block address of new partition. Adding a check condition to avoid expanding partition in safe mode, and change manual accordingly. Signed-off-by: Chao Yu <chao@kernel.org> --- fsck/resize.c | 12 ++++++++---- man/resize.f2fs.8 | 2 +- 2 files changed, 9 insertions(+), 5 deletions(-)