2019-01-29 17:51:58

by Artem Blagodarenko

[permalink] [raw]
Subject: [PATCH v7 0/4] e2image -b option to pass superblock number

There was a discussion about possibility of using
superblock backup for creating images with e2image.

https://www.spinics.net/lists/linux-ext4/msg30583.html

Now our customer faced with situation then it would be
useful to have image file from partition with broken
superblock.

Here is set of patches that adds this functionality.

This is 7th version of the patch set.

Changes from previous version:

* Only two cases in try_open_fs() now: with and without passed
superblock size.
* Some minor code style and comments issues are fixed.

Artem Blagodarenko (4):
ext2fs: opening filesystem code refactoring
e2image: add -b and -B options to use supperblock backup
tests: add test for e2image -b option
ext2fs: automaticlly open backup superblocks

e2fsck/e2fsck.h | 2 -
e2fsck/message.c | 3 +-
e2fsck/unix.c | 42 ++++---------
e2fsck/util.c | 73 -----------------------
lib/ext2fs/openfs.c | 39 ++++++++++--
lib/support/Makefile.in | 8 ++-
lib/support/sb_backup.c | 140 ++++++++++++++++++++++++++++++++++++++++++++
lib/support/sb_backup.h | 20 +++++++
misc/dumpe2fs.c | 17 +-----
misc/e2image.8.in | 33 +++++++++++
misc/e2image.c | 50 ++++++++++++++--
tests/i_zero_super/expect.1 | 22 +++++++
tests/i_zero_super/image.gz | Bin 0 -> 13262 bytes
tests/i_zero_super/script | 11 ++++
14 files changed, 328 insertions(+), 132 deletions(-)
create mode 100644 lib/support/sb_backup.c
create mode 100644 lib/support/sb_backup.h
create mode 100644 tests/i_zero_super/expect.1
create mode 100644 tests/i_zero_super/image.gz
create mode 100644 tests/i_zero_super/script

--
2.14.3



2019-01-29 17:52:00

by Artem Blagodarenko

[permalink] [raw]
Subject: [PATCH v7 1/4] ext2fs: opening filesystem code refactoring

There are similar opening filesystem code in different utilities.

The patch moves improved handling from try_open_fs()
into ext2fs_open(). This function make one of the action
based on parameters:
1) open filesystem with given superblock, superblock size
2) open filesystem with given superblock, but try to
find right block size

Signed-off-by: Artem Blagodarenko <[email protected]>
---
e2fsck/unix.c | 28 +++-------------------------
lib/ext2fs/openfs.c | 39 +++++++++++++++++++++++++++++++++++----
misc/dumpe2fs.c | 17 ++---------------
3 files changed, 40 insertions(+), 44 deletions(-)

diff --git a/e2fsck/unix.c b/e2fsck/unix.c
index 5b3552ec..ddcf52a4 100644
--- a/e2fsck/unix.c
+++ b/e2fsck/unix.c
@@ -1151,31 +1151,9 @@ static errcode_t try_open_fs(e2fsck_t ctx, int flags, io_manager io_ptr,
ext2_filsys *ret_fs)
{
errcode_t retval;
-
- *ret_fs = NULL;
- if (ctx->superblock && ctx->blocksize) {
- retval = ext2fs_open2(ctx->filesystem_name, ctx->io_options,
- flags, ctx->superblock, ctx->blocksize,
- io_ptr, ret_fs);
- } else if (ctx->superblock) {
- int blocksize;
- for (blocksize = EXT2_MIN_BLOCK_SIZE;
- blocksize <= EXT2_MAX_BLOCK_SIZE; blocksize *= 2) {
- if (*ret_fs) {
- ext2fs_free(*ret_fs);
- *ret_fs = NULL;
- }
- retval = ext2fs_open2(ctx->filesystem_name,
- ctx->io_options, flags,
- ctx->superblock, blocksize,
- io_ptr, ret_fs);
- if (!retval)
- break;
- }
- } else
- retval = ext2fs_open2(ctx->filesystem_name, ctx->io_options,
- flags, 0, 0, io_ptr, ret_fs);
-
+ retval = ext2fs_open2(ctx->filesystem_name, ctx->io_options,
+ flags, ctx->superblock, ctx->blocksize,
+ io_ptr, ret_fs);
if (retval == 0) {
(*ret_fs)->priv_data = ctx;
e2fsck_set_bitmap_type(*ret_fs, EXT2FS_BMAP64_RBTREE,
diff --git a/lib/ext2fs/openfs.c b/lib/ext2fs/openfs.c
index 85d73e2a..16d36200 100644
--- a/lib/ext2fs/openfs.c
+++ b/lib/ext2fs/openfs.c
@@ -114,10 +114,10 @@ static void block_sha_map_free_entry(void *data)
* EXT2_FLAG_64BITS - Allow 64-bit bitfields (needed for large
* filesystems)
*/
-errcode_t ext2fs_open2(const char *name, const char *io_options,
- int flags, int superblock,
- unsigned int block_size, io_manager manager,
- ext2_filsys *ret_fs)
+static errcode_t __ext2fs_open2(const char *name, const char *io_options,
+ int flags, int superblock,
+ unsigned int block_size, io_manager manager,
+ ext2_filsys *ret_fs)
{
ext2_filsys fs;
errcode_t retval;
@@ -515,6 +515,37 @@ cleanup:
return retval;
}

+errcode_t ext2fs_open2(const char *name, const char *io_options,
+ int flags, int superblock,
+ unsigned int block_size, io_manager manager,
+ ext2_filsys *ret_fs)
+{
+ errcode_t retval;
+
+ *ret_fs = NULL;
+ if (superblock && block_size) {
+ retval = __ext2fs_open2(name, io_options,
+ flags, superblock, block_size,
+ manager, ret_fs);
+ } else {
+ int size;
+
+ for (size = EXT2_MIN_BLOCK_SIZE;
+ size <= EXT2_MAX_BLOCK_SIZE; size *= 2) {
+ if (*ret_fs) {
+ ext2fs_free(*ret_fs);
+ *ret_fs = NULL;
+ }
+ retval = __ext2fs_open2(name, io_options, flags,
+ superblock, size,
+ manager, ret_fs);
+ if (!retval)
+ break;
+ }
+ }
+ return retval;
+}
+
/*
* Set/get the filesystem data I/O channel.
*
diff --git a/misc/dumpe2fs.c b/misc/dumpe2fs.c
index 384ce925..183e2f6f 100644
--- a/misc/dumpe2fs.c
+++ b/misc/dumpe2fs.c
@@ -667,21 +667,8 @@ int main (int argc, char ** argv)
if (image_dump)
flags |= EXT2_FLAG_IMAGE_FILE;
try_open_again:
- if (use_superblock && !use_blocksize) {
- for (use_blocksize = EXT2_MIN_BLOCK_SIZE;
- use_blocksize <= EXT2_MAX_BLOCK_SIZE;
- use_blocksize *= 2) {
- retval = ext2fs_open (device_name, flags,
- use_superblock,
- use_blocksize, unix_io_manager,
- &fs);
- if (!retval)
- break;
- }
- } else {
- retval = ext2fs_open(device_name, flags, use_superblock,
- use_blocksize, unix_io_manager, &fs);
- }
+ retval = ext2fs_open2(device_name, 0, flags, use_superblock,
+ use_blocksize, unix_io_manager, &fs);
flags |= EXT2_FLAG_IGNORE_CSUM_ERRORS;
if (retval && !retval_csum) {
retval_csum = retval;
--
2.14.3


2019-01-29 17:52:02

by Artem Blagodarenko

[permalink] [raw]
Subject: [PATCH v7 2/4] e2image: add -b and -B options to use supperblock backup

e2image has no ability to use superblock backup to copy metadata.
This feature can be useful if someone wants to make partition
image and fix it using e2fsck utility.

New -b option allows to pass superblock number, like e2fsck utility does.
e2image doesn't change primary superblock and store it as is, so
it can be fixed using e2fsck latter. Option -B allows setting
superblock size.

Signed-off-by: Artem Blagodarenko <[email protected]>
---
misc/e2image.8.in | 33 +++++++++++++++++++++++++++++++++
misc/e2image.c | 43 +++++++++++++++++++++++++++++++++++++------
2 files changed, 70 insertions(+), 6 deletions(-)

diff --git a/misc/e2image.8.in b/misc/e2image.8.in
index a7bfdf24..bbbb57ae 100644
--- a/misc/e2image.8.in
+++ b/misc/e2image.8.in
@@ -12,6 +12,15 @@ e2image \- Save critical ext2/ext3/ext4 filesystem metadata to a file
]
[
.B \-f
+.B \-b
+.I superblock
+]
+[
+.B \-B
+.I blocksize
+]
+[
+.B \-fr
]
.I device
.I image-file
@@ -167,6 +176,22 @@ the
option will prevent analysis of problems related to hash-tree indexed
directories.
.PP
+Option
+.B \-b
+.I superblock
+can be used to get image from partition with broken primary superblock.
+The partition is copied as-is including broken primary superblock.
+.PP
+Option
+.B \-B
+.I blocksize
+can be used to set superblock block size. Normally, e2fsck will search
+for the superblock at various different block sizes in an attempt to find
+the appropriate blocksize. This search can be fooled in some cases. This
+option forces e2fsck to only try locating the superblock at a particular
+blocksize. If the superblock is not found, e2fsck will terminate with a
+fatal error.
+.PP
Note that this will work even if you substitute "/dev/hda1" for another raw
disk image, or QCOW2 image previously created by
.BR e2image .
@@ -217,6 +242,14 @@ This can be useful to write a qcow2 image containing all data to a
sparse image file where it can be loop mounted, or to a disk partition.
Note that this may not work with qcow2 images not generated by e2image.
.PP
+Options
+.B \-b
+.I superblock
+and
+.B \-B
+.I blocksize
+can be used same way as for raw images.
+.PP
.SH INCLUDING DATA
Normally
.B e2image
diff --git a/misc/e2image.c b/misc/e2image.c
index 9e21d0db..3c881fee 100644
--- a/misc/e2image.c
+++ b/misc/e2image.c
@@ -104,7 +104,8 @@ static int get_bits_from_size(size_t size)

static void usage(void)
{
- fprintf(stderr, _("Usage: %s [ -r|Q ] [ -f ] device image-file\n"),
+ fprintf(stderr, _("Usage: %s [ -r|Q ] [ -f ] [ -b superblock ] [ -B blocksize]"
+ "[ -fr ] device image-file\n"),
program_name);
fprintf(stderr, _(" %s -I device image-file\n"), program_name);
fprintf(stderr, _(" %s -ra [ -cfnp ] [ -o src_offset ] "
@@ -1267,7 +1268,8 @@ static void output_qcow2_meta_data_blocks(ext2_filsys fs, int fd)
free_qcow2_image(img);
}

-static void write_raw_image_file(ext2_filsys fs, int fd, int type, int flags)
+static void write_raw_image_file(ext2_filsys fs, int fd, int type, int flags,
+ blk64_t superblock)
{
struct process_block_struct pb;
struct ext2_inode inode;
@@ -1295,6 +1297,22 @@ static void write_raw_image_file(ext2_filsys fs, int fd, int type, int flags)
}
}

+ if (superblock) {
+ int j;
+
+ ext2fs_mark_block_bitmap2(meta_block_map, superblock);
+ meta_blocks_count++;
+
+ /*
+ * Mark the backup superblock descriptors
+ */
+ for (j = 0; j < fs->desc_blocks; j++) {
+ ext2fs_mark_block_bitmap2(meta_block_map,
+ ext2fs_descriptor_block_loc2(fs, superblock, j));
+ }
+ meta_blocks_count += fs->desc_blocks;
+ }
+
mark_table_blocks(fs);
if (show_progress)
fprintf(stderr, "%s", _("Scanning inodes...\n"));
@@ -1474,6 +1492,8 @@ int main (int argc, char ** argv)
int ignore_rw_mount = 0;
int check = 0;
struct stat st;
+ blk64_t superblock = 0;
+ int blocksize = 0;

#ifdef ENABLE_NLS
setlocale(LC_MESSAGES, "");
@@ -1487,8 +1507,14 @@ int main (int argc, char ** argv)
if (argc && *argv)
program_name = *argv;
add_error_table(&et_ext2_error_table);
- while ((c = getopt(argc, argv, "nrsIQafo:O:pc")) != EOF)
+ while ((c = getopt(argc, argv, "b:B:nrsIQafo:O:pc")) != EOF)
switch (c) {
+ case 'b':
+ superblock = strtoull(optarg, NULL, 0);
+ break;
+ case 'B':
+ blocksize = strtoul(optarg, NULL, 0);
+ break;
case 'I':
flags |= E2IMAGE_INSTALL_FLAG;
break;
@@ -1540,6 +1566,11 @@ int main (int argc, char ** argv)
"with raw or QCOW2 images."));
exit(1);
}
+ if (superblock && !img_type) {
+ com_err(program_name, 0, "%s", _("-b option can only be used "
+ "with raw or QCOW2 images."));
+ exit(1);
+ }
if ((source_offset || dest_offset) && img_type != E2IMAGE_RAW) {
com_err(program_name, 0, "%s",
_("Offsets are only allowed with raw images."));
@@ -1590,8 +1621,8 @@ int main (int argc, char ** argv)
}
}
sprintf(offset_opt, "offset=%llu", source_offset);
- retval = ext2fs_open2(device_name, offset_opt, open_flag, 0, 0,
- unix_io_manager, &fs);
+ retval = ext2fs_open2(device_name, offset_opt, open_flag,
+ superblock, blocksize, unix_io_manager, &fs);
if (retval) {
com_err (program_name, retval, _("while trying to open %s"),
device_name);
@@ -1681,7 +1712,7 @@ skip_device:
exit(1);
}
if (img_type)
- write_raw_image_file(fs, fd, img_type, flags);
+ write_raw_image_file(fs, fd, img_type, flags, superblock);
else
write_image_file(fs, fd);

--
2.14.3


2019-01-29 17:52:04

by Artem Blagodarenko

[permalink] [raw]
Subject: [PATCH v7 3/4] tests: add test for e2image -b option

The test makes raw image from partition with broken super block
and executes e2fsck.

Signed-off-by: Artem Blagodarenko <[email protected]>
---
tests/i_zero_super/expect.1 | 22 ++++++++++++++++++++++
tests/i_zero_super/image.gz | Bin 0 -> 13262 bytes
tests/i_zero_super/script | 11 +++++++++++
3 files changed, 33 insertions(+)

diff --git a/tests/i_zero_super/expect.1 b/tests/i_zero_super/expect.1
new file mode 100644
index 00000000..fda32b96
--- /dev/null
+++ b/tests/i_zero_super/expect.1
@@ -0,0 +1,22 @@
+Pass 1: Checking inodes, blocks, and sizes
+Pass 2: Checking directory structure
+Pass 3: Checking directory connectivity
+Pass 4: Checking reference counts
+Pass 5: Checking group summary information
+Free blocks count wrong for group #0 (7987, counted=7982).
+Fix? no
+
+Free blocks count wrong (11602, counted=11597).
+Fix? no
+
+Free inodes count wrong for group #0 (1493, counted=1488).
+Fix? no
+
+Free inodes count wrong (2997, counted=2992).
+Fix? no
+
+
+test_filesys: ********** WARNING: Filesystem still has errors **********
+
+test_filesys: 11/3008 files (0.0% non-contiguous), 398/12000 blocks
+Exit status is 4
diff --git a/tests/i_zero_super/image.gz b/tests/i_zero_super/image.gz
new file mode 100644
index 0000000000000000000000000000000000000000..eea9140194f4dd9ec9dc8bb7b9e2215b950587d2
GIT binary patch
literal 13262
zcmeI&{WsHl902f@Tih#ASFE<IL%5VllFHLfbsepcr#yxVS=!QEX3V<Xij3TH)g@2m
zDS0T_=4nglT7;E{xfz>0G_ld9Vc5R+bpL^V>zwZSynpzd^M0St`F!5z^~3w8H;SqT
zqYs|hy;?=Ydxb*`Y=(_Uj;z^->s>9@r^5EDn{U+Cn|rW>b}kfCoqXW-K~s(F;j74#
zhqt?LGco2DyY?LC#w8Uk#19^?CI2?lfagqx1)ifG8HH6hag-M>b#n;a0ZPl9H5uQ*
zh;IX^QFd*|QbjWJu&O+~X7%S}AV@xtYOvP)CS>1lrK);rafz~F03E!AP48O_FHJ-S
zg=;|Z6CYQNv-ZtIAsw=E5_ROA?i#k~veE+)ipFdv$#ixO-jpJlhBw`i@Ze2U$ugXh
zEMdbbDH0)^B6nzm#&w;gg^SRCFE9f3*C-V-jA$$~oJK~np*>kNW!_c@&xu+YHyNXN
zp%GA9=BovWD%_K}kqu4M8l_Vdo`ZrjD`Rn~ny_y)f?E3F0+*`oZ`-HKopl65$&V?2
z6t#B;?@_*MZBA_RXBf6G+|DYp-XhkO$YO6^T9V`5bn#K^+MJ~w%?yMSFJ^6Sg0UPo
zH18)>^zs;IYTrH~YA<7Mr$64hU|~C{vPc@*cM^i?V|<|Jn9+tquql)qba_d&&!IrM
zN~Zac^0PwabLqMD!9ml85Ia=TGE1p+3h@zX%O9$xRb*SWAT&w23469Mvy?^fd1|g3
z2OBk4fRetx=bz*1p{n()`A%z>On+9V^<KoFScBS4+?9aD>5jQ%qmv|KEd9g7%{cya
z65*HM6(OZ+6L>k!K_u1|@&p_klxX(aQxoxcuK2-5NAEJvycMqbFk(JtZeOq(eaVqK
zE$C<*4!NVbAnmvsr0$0sDX)ZlCMriySRgmmUjO^tlevd#3)Zt$LqUQ_eDSM@rLm)g
z!XAD=Sx#gZABPJ)Up{;GKD|C}Y8VkwAwFuhGbzSGG-0EPN6htOi;<%tBC<{2D7M<c
zHgZ1}&X(q@H%g}2b?~jUzVP!H8P+O>Hvi^SK!zjPB6^PPltypW6U@ls8<!hHlm+Zo
zhV@ue6ayjtKI3?=--?Te84ndnH!`e)6ZcO$@NXO<7^J3z)N8$XWIo%2ak^=WHQZ6I
z@Q$yP_0qj<cs{(29;$C_Nk@Z?7kA3j_q<0a%gT$(DtpQ;?ZCRc%)YpelMq?#?(f$b
z_UamTFN+X&iEQDYQcs_9B*mswP5Tz!>I$Qi8V+~~sMW&Avh1qk{jsdUms4q%_I~j9
zAP1wBH!x%p{*lPf%!OjFcUM;S-X3L~(Ph>1KAIhBU*6R=9GDzLN>@M4QqZpW;86in
zU6u2kydgWZ5!3%Mj+USKnvbny#tWU>-s*8m*m3AEb(yr<){ZABVq)XGI#2cz42~T&
zGiMA2h%d5h8S#Z$(EIzA$Y}eQPcMd3koG%AZ~YY0O4M39@<xLdrBh?hKWLtF{hBCP
zKSKVfKI(kvuk_4HnqI|OAvLr|+>qm%pxYt5T=-)((!SH_-N)wOT5EJu-k$_?OS4By
z`-cNru4=m6(i&s&<}6`1-*Pu6zm4e05s8Xx>Q+>QMg+NP$urt|PT#~`t~3AhF_nSY
zvBHEWF`>oD!xz}9ttUu?ntrBkM$M-%!q|)=)_Y)mo((>q^l>hC-8VW!K-)k72mk>f
z00e*l5C8%|00;m9AOHk_01yBIKmZ5;0U!VbfB+Bx0zd!=00AKIbp=K!;yw%P>JwUP
zN8Z$BYH3yUJBnz@L=EacMGN|;r>&TdDbac%hz*WE#2d70bXsDzshu_)m6AW|wjzkT
zcBy>bZol}OTTD6WMU0ljJ&icy#4Kr$Hr@9)dvDXkMN4wZxwi50$>mD+G8Nh><%Z8z
z-r_hsyIZ&rI57ThUe6tC!j|bV27G0o?8J#?hL{KkW#FoUb?d|b>Kwkeb>zx!KDuCK
v`J3J9`63;-00e*l5C8%|00;m9AOHk_01yBI{~LklvK5${_ikI*<wn@Q%S-lK

literal 0
HcmV?d00001

diff --git a/tests/i_zero_super/script b/tests/i_zero_super/script
new file mode 100644
index 00000000..6b02d86d
--- /dev/null
+++ b/tests/i_zero_super/script
@@ -0,0 +1,11 @@
+if test -x $E2IMAGE_EXE; then
+
+ONE_PASS_ONLY=true
+FSCK_OPT="-n -b 8193 -f"
+PREP_CMD="$E2IMAGE_EXE -r -b 8193 -B 1024 $TMPFILE $TMPFILE.raw; \
+ mv $TMPFILE.raw $TMPFILE"
+. $cmd_dir/run_e2fsck
+
+else
+ echo "$test_name: $test_description: skipped"
+fi
--
2.14.3


2019-01-29 17:52:07

by Artem Blagodarenko

[permalink] [raw]
Subject: [PATCH v7 4/4] ext2fs: automaticlly open backup superblocks

e2image and e2fsck automatically try to open some backup superblocks,
if only blocksize is set or passed superblock can't be opened.
Try few backup superblocks (e.g. {1, 3, 5, 7, 9} * blocksize * 8).

This code is moved to lib/support/.

Signed-off-by: Artem Blagodarenko <[email protected]>
---
e2fsck/e2fsck.h | 2 -
e2fsck/message.c | 3 +-
e2fsck/unix.c | 14 +++--
e2fsck/util.c | 73 -------------------------
lib/support/Makefile.in | 8 ++-
lib/support/sb_backup.c | 140 ++++++++++++++++++++++++++++++++++++++++++++++++
lib/support/sb_backup.h | 20 +++++++
misc/e2image.c | 7 +++
8 files changed, 185 insertions(+), 82 deletions(-)

diff --git a/e2fsck/e2fsck.h b/e2fsck/e2fsck.h
index 1c7a67cb..8576ef24 100644
--- a/e2fsck/e2fsck.h
+++ b/e2fsck/e2fsck.h
@@ -633,8 +633,6 @@ extern void e2fsck_write_inode_full(e2fsck_t ctx, unsigned long ino,
#ifdef MTRACE
extern void mtrace_print(char *mesg);
#endif
-extern blk64_t get_backup_sb(e2fsck_t ctx, ext2_filsys fs,
- const char *name, io_manager manager);
extern int ext2_file_type(unsigned int mode);
extern int write_all(int fd, char *buf, size_t count);
void dump_mmp_msg(struct mmp_struct *mmp, const char *fmt, ...)
diff --git a/e2fsck/message.c b/e2fsck/message.c
index 727f71d5..ffb40730 100644
--- a/e2fsck/message.c
+++ b/e2fsck/message.c
@@ -465,7 +465,8 @@ static _INLINE_ void expand_percent_expression(FILE *f, ext2_filsys fs,
fprintf(f, "%*lld", width, (long long) ctx->blkcount);
break;
case 'S':
- fprintf(f, "%llu", get_backup_sb(NULL, fs, NULL, NULL));
+ fprintf(f, "%llu", get_first_backup_sb(NULL, NULL, fs,
+ NULL, NULL));
break;
case 's':
fprintf(f, "%*s", width, ctx->str ? ctx->str : "NULL");
diff --git a/e2fsck/unix.c b/e2fsck/unix.c
index ddcf52a4..4ffc039e 100644
--- a/e2fsck/unix.c
+++ b/e2fsck/unix.c
@@ -1489,11 +1489,17 @@ restart:
retval ? _("Superblock invalid,") :
_("Group descriptors look bad..."));
orig_superblock = ctx->superblock;
- get_backup_sb(ctx, fs, ctx->filesystem_name, io_ptr);
- if (fs)
- ext2fs_close_free(&fs);
orig_retval = retval;
- retval = try_open_fs(ctx, flags, io_ptr, &fs);
+ retval = try_backups(ctx->filesystem_name,
+ ctx->io_options,
+ flags, &ctx->superblock,
+ &ctx->blocksize, io_ptr, &fs);
+ if (retval == 0) {
+ fs->priv_data = ctx;
+ e2fsck_set_bitmap_type(fs,
+ EXT2FS_BMAP64_RBTREE,
+ "default", NULL);
+ }
if ((orig_retval == 0) && retval != 0) {
if (fs)
ext2fs_close_free(&fs);
diff --git a/e2fsck/util.c b/e2fsck/util.c
index db6a1cc1..e2caf3a2 100644
--- a/e2fsck/util.c
+++ b/e2fsck/util.c
@@ -549,79 +549,6 @@ void mtrace_print(char *mesg)
}
#endif

-blk64_t get_backup_sb(e2fsck_t ctx, ext2_filsys fs, const char *name,
- io_manager manager)
-{
- struct ext2_super_block *sb;
- io_channel io = NULL;
- void *buf = NULL;
- int blocksize;
- blk64_t superblock, ret_sb = 8193;
-
- if (fs && fs->super) {
- ret_sb = (fs->super->s_blocks_per_group +
- fs->super->s_first_data_block);
- if (ctx) {
- ctx->superblock = ret_sb;
- ctx->blocksize = fs->blocksize;
- }
- return ret_sb;
- }
-
- if (ctx) {
- if (ctx->blocksize) {
- ret_sb = ctx->blocksize * 8;
- if (ctx->blocksize == 1024)
- ret_sb++;
- ctx->superblock = ret_sb;
- return ret_sb;
- }
- ctx->superblock = ret_sb;
- ctx->blocksize = 1024;
- }
-
- if (!name || !manager)
- goto cleanup;
-
- if (manager->open(name, 0, &io) != 0)
- goto cleanup;
-
- if (ext2fs_get_mem(SUPERBLOCK_SIZE, &buf))
- goto cleanup;
- sb = (struct ext2_super_block *) buf;
-
- for (blocksize = EXT2_MIN_BLOCK_SIZE;
- blocksize <= EXT2_MAX_BLOCK_SIZE ; blocksize *= 2) {
- superblock = blocksize*8;
- if (blocksize == 1024)
- superblock++;
- io_channel_set_blksize(io, blocksize);
- if (io_channel_read_blk64(io, superblock,
- -SUPERBLOCK_SIZE, buf))
- continue;
-#ifdef WORDS_BIGENDIAN
- if (sb->s_magic == ext2fs_swab16(EXT2_SUPER_MAGIC))
- ext2fs_swap_super(sb);
-#endif
- if ((sb->s_magic == EXT2_SUPER_MAGIC) &&
- (EXT2_BLOCK_SIZE(sb) == blocksize)) {
- ret_sb = superblock;
- if (ctx) {
- ctx->superblock = superblock;
- ctx->blocksize = blocksize;
- }
- break;
- }
- }
-
-cleanup:
- if (io)
- io_channel_close(io);
- if (buf)
- ext2fs_free_mem(&buf);
- return (ret_sb);
-}
-
/*
* Given a mode, return the ext2 file type
*/
diff --git a/lib/support/Makefile.in b/lib/support/Makefile.in
index 1d278642..03fdf555 100644
--- a/lib/support/Makefile.in
+++ b/lib/support/Makefile.in
@@ -22,7 +22,8 @@ OBJS= cstring.o \
quotaio.o \
quotaio_v2.o \
quotaio_tree.o \
- dict.o
+ dict.o \
+ sb_backup.o

SRCS= $(srcdir)/argv_parse.c \
$(srcdir)/cstring.c \
@@ -35,7 +36,8 @@ SRCS= $(srcdir)/argv_parse.c \
$(srcdir)/quotaio.c \
$(srcdir)/quotaio_tree.c \
$(srcdir)/quotaio_v2.c \
- $(srcdir)/dict.c
+ $(srcdir)/dict.c \
+ $(srcdir)/sb_backup.c

LIBRARY= libsupport
LIBDIR= support
@@ -168,3 +170,5 @@ quotaio_v2.o: $(srcdir)/quotaio_v2.c $(top_builddir)/lib/config.h \
$(srcdir)/quotaio_tree.h
dict.o: $(srcdir)/dict.c $(top_builddir)/lib/config.h \
$(top_builddir)/lib/dirpaths.h $(srcdir)/dict.h
+sb_backup.o: $(srcdir)/sb_backup.c $(top_builddir)/lib/config.h \
+ $(srcdir)/sb_backup.h
diff --git a/lib/support/sb_backup.c b/lib/support/sb_backup.c
new file mode 100644
index 00000000..cfa671e5
--- /dev/null
+++ b/lib/support/sb_backup.c
@@ -0,0 +1,140 @@
+/*
+ * sb_backup.c -- helper functions for getting backup superblocks
+ *
+ * %Begin-Header%
+ * This file may be redistributed under the terms of the GNU Library
+ * General Public License, version 2.
+ * %End-Header%
+ */
+
+
+#include "config.h"
+#include <stdio.h>
+#include <string.h>
+
+#include "ext2fs/ext2_fs.h"
+#include "ext2fs/ext2fs.h"
+
+blk64_t get_first_backup_sb(blk64_t *superblock, unsigned int *block_size,
+ ext2_filsys fs, const char *name,
+ io_manager manager)
+{
+ struct ext2_super_block *sb;
+ io_channel io = NULL;
+ void *buf = NULL;
+ int try_blocksize;
+ blk64_t try_superblock, ret_sb = 8193;
+
+ /* superblock and block_size can be NULL if fs->super is passed */
+ if (fs && fs->super) {
+ ret_sb = fs->super->s_blocks_per_group +
+ fs->super->s_first_data_block;
+ if (superblock)
+ *superblock = ret_sb;
+ if (block_size)
+ *block_size = fs->blocksize;
+ return ret_sb;
+ }
+
+ if (*block_size) {
+ ret_sb = *block_size * 8;
+ if (*block_size == 1024)
+ ret_sb++;
+ *superblock = ret_sb;
+ return ret_sb;
+ }
+
+ *superblock = ret_sb;
+ *block_size = 1024;
+
+ if (!name || !manager)
+ goto cleanup;
+
+ if (manager->open(name, 0, &io) != 0)
+ goto cleanup;
+
+ if (ext2fs_get_mem(SUPERBLOCK_SIZE, &buf))
+ goto cleanup;
+ sb = (struct ext2_super_block *) buf;
+
+ for (try_blocksize = EXT2_MIN_BLOCK_SIZE;
+ try_blocksize <= EXT2_MAX_BLOCK_SIZE ; try_blocksize *= 2) {
+ try_superblock = try_blocksize*8;
+ if (try_blocksize == 1024)
+ try_superblock++;
+ io_channel_set_blksize(io, try_blocksize);
+ if (io_channel_read_blk64(io, try_superblock,
+ -SUPERBLOCK_SIZE, buf))
+ continue;
+#ifdef WORDS_BIGENDIAN
+ if (sb->s_magic == ext2fs_swab16(EXT2_SUPER_MAGIC))
+ ext2fs_swap_super(sb);
+#endif
+ if ((sb->s_magic == EXT2_SUPER_MAGIC) &&
+ (EXT2_BLOCK_SIZE(sb) == try_blocksize)) {
+ ret_sb = try_superblock;
+ *superblock = try_superblock;
+ *block_size = try_blocksize;
+ break;
+ }
+ }
+
+cleanup:
+ if (io)
+ io_channel_close(io);
+ if (buf)
+ ext2fs_free_mem(&buf);
+ return ret_sb;
+}
+
+errcode_t try_backups(const char *name, const char *io_options,
+ int flags, blk64_t *superblock,
+ unsigned int *block_size, io_manager manager,
+ ext2_filsys *ret_fs)
+{
+ errcode_t retval, retval2;
+ blk64_t try_block_number;
+ unsigned int i;
+
+ /*
+ * Get first superblock location based on heuristic.
+ * Blocksize is also returned and used to find next
+ * superblock copy location.
+ */
+ try_block_number = get_first_backup_sb(superblock, block_size,
+ *ret_fs, name, manager);
+ retval = ext2fs_open2(name, io_options, flags, try_block_number,
+ *block_size, manager, ret_fs);
+ if (!retval)
+ return 0;
+
+ /*
+ * Try 3d, 5th, 7th, 9th superblock copy
+ */
+ for (i = 3; i <= 9; i += 2) {
+ try_block_number = (i * (*block_size) * 8);
+ if (*block_size == 1024)
+ try_block_number++;
+ if (*ret_fs) {
+ ext2fs_free(*ret_fs);
+ *ret_fs = NULL;
+ }
+ retval2 = ext2fs_open2(name, io_options, flags,
+ try_block_number, *block_size, manager,
+ ret_fs);
+ /*
+ * Partition is too small to have this superblock copy.
+ * Return previous reading return code
+ */
+ if (retval2 == EXT2_ET_SHORT_READ)
+ break;
+
+ retval = retval2;
+ if (!retval) {
+ *superblock = try_block_number;
+ break;
+ }
+ }
+
+ return retval;
+}
diff --git a/lib/support/sb_backup.h b/lib/support/sb_backup.h
new file mode 100644
index 00000000..1e18e37c
--- /dev/null
+++ b/lib/support/sb_backup.h
@@ -0,0 +1,20 @@
+/*
+ * sb_backup.h -- helper functions for getting backup superblocks
+ *
+ * %Begin-Header%
+ * This file may be redistributed under the terms of the GNU Library
+ * General Public License, version 2.
+ * %End-Header%
+ */
+
+#include "ext2fs/ext2_fs.h"
+#include "ext2fs/ext2fs.h"
+
+blk64_t get_first_backup_sb(blk64_t *superblock, unsigned int *block_size,
+ ext2_filsys fs, const char *name,
+ io_manager manager);
+
+errcode_t try_backups(const char *name, const char *io_options,
+ int flags, blk64_t *superblock,
+ unsigned int *block_size, io_manager manager,
+ ext2_filsys *ret_fs);
diff --git a/misc/e2image.c b/misc/e2image.c
index 3c881fee..786282ec 100644
--- a/misc/e2image.c
+++ b/misc/e2image.c
@@ -1623,6 +1623,13 @@ int main (int argc, char ** argv)
sprintf(offset_opt, "offset=%llu", source_offset);
retval = ext2fs_open2(device_name, offset_opt, open_flag,
superblock, blocksize, unix_io_manager, &fs);
+ if (retval & (superblock | blocksize)) {
+ printf(_("Try backups in other location.\n"));
+ retval = try_backups(device_name, offset_opt, open_flag,
+ &superblock, &blocksize,
+ unix_io_manager, &fs);
+ printf(_("Use superblock %i.\n"), superblock);
+ }
if (retval) {
com_err (program_name, retval, _("while trying to open %s"),
device_name);
--
2.14.3


2019-01-30 03:01:09

by Andreas Dilger

[permalink] [raw]
Subject: Re: [PATCH v7 1/4] ext2fs: opening filesystem code refactoring

On Jan 29, 2019, at 10:51 AM, Artem Blagodarenko <[email protected]> wrote:
>
> There are similar opening filesystem code in different utilities.
>
> The patch moves improved handling from try_open_fs()
> into ext2fs_open(). This function make one of the action
> based on parameters:
> 1) open filesystem with given superblock, superblock size
> 2) open filesystem with given superblock, but try to
> find right block size
>
> Signed-off-by: Artem Blagodarenko <[email protected]>

Reviewed-by: Andreas Dilger <[email protected]>

> ---
> e2fsck/unix.c | 28 +++-------------------------
> lib/ext2fs/openfs.c | 39 +++++++++++++++++++++++++++++++++++----
> misc/dumpe2fs.c | 17 ++---------------
> 3 files changed, 40 insertions(+), 44 deletions(-)
>
> diff --git a/e2fsck/unix.c b/e2fsck/unix.c
> index 5b3552ec..ddcf52a4 100644
> --- a/e2fsck/unix.c
> +++ b/e2fsck/unix.c
> @@ -1151,31 +1151,9 @@ static errcode_t try_open_fs(e2fsck_t ctx, int flags, io_manager io_ptr,
> ext2_filsys *ret_fs)
> {
> errcode_t retval;
> -
> - *ret_fs = NULL;
> - if (ctx->superblock && ctx->blocksize) {
> - retval = ext2fs_open2(ctx->filesystem_name, ctx->io_options,
> - flags, ctx->superblock, ctx->blocksize,
> - io_ptr, ret_fs);
> - } else if (ctx->superblock) {
> - int blocksize;
> - for (blocksize = EXT2_MIN_BLOCK_SIZE;
> - blocksize <= EXT2_MAX_BLOCK_SIZE; blocksize *= 2) {
> - if (*ret_fs) {
> - ext2fs_free(*ret_fs);
> - *ret_fs = NULL;
> - }
> - retval = ext2fs_open2(ctx->filesystem_name,
> - ctx->io_options, flags,
> - ctx->superblock, blocksize,
> - io_ptr, ret_fs);
> - if (!retval)
> - break;
> - }
> - } else
> - retval = ext2fs_open2(ctx->filesystem_name, ctx->io_options,
> - flags, 0, 0, io_ptr, ret_fs);
> -
> + retval = ext2fs_open2(ctx->filesystem_name, ctx->io_options,
> + flags, ctx->superblock, ctx->blocksize,
> + io_ptr, ret_fs);
> if (retval == 0) {
> (*ret_fs)->priv_data = ctx;
> e2fsck_set_bitmap_type(*ret_fs, EXT2FS_BMAP64_RBTREE,
> diff --git a/lib/ext2fs/openfs.c b/lib/ext2fs/openfs.c
> index 85d73e2a..16d36200 100644
> --- a/lib/ext2fs/openfs.c
> +++ b/lib/ext2fs/openfs.c
> @@ -114,10 +114,10 @@ static void block_sha_map_free_entry(void *data)
> * EXT2_FLAG_64BITS - Allow 64-bit bitfields (needed for large
> * filesystems)
> */
> -errcode_t ext2fs_open2(const char *name, const char *io_options,
> - int flags, int superblock,
> - unsigned int block_size, io_manager manager,
> - ext2_filsys *ret_fs)
> +static errcode_t __ext2fs_open2(const char *name, const char *io_options,
> + int flags, int superblock,
> + unsigned int block_size, io_manager manager,
> + ext2_filsys *ret_fs)
> {
> ext2_filsys fs;
> errcode_t retval;
> @@ -515,6 +515,37 @@ cleanup:
> return retval;
> }
>
> +errcode_t ext2fs_open2(const char *name, const char *io_options,
> + int flags, int superblock,
> + unsigned int block_size, io_manager manager,
> + ext2_filsys *ret_fs)
> +{
> + errcode_t retval;
> +
> + *ret_fs = NULL;
> + if (superblock && block_size) {
> + retval = __ext2fs_open2(name, io_options,
> + flags, superblock, block_size,
> + manager, ret_fs);
> + } else {
> + int size;
> +
> + for (size = EXT2_MIN_BLOCK_SIZE;
> + size <= EXT2_MAX_BLOCK_SIZE; size *= 2) {
> + if (*ret_fs) {
> + ext2fs_free(*ret_fs);
> + *ret_fs = NULL;
> + }
> + retval = __ext2fs_open2(name, io_options, flags,
> + superblock, size,
> + manager, ret_fs);
> + if (!retval)
> + break;
> + }
> + }
> + return retval;
> +}
> +
> /*
> * Set/get the filesystem data I/O channel.
> *
> diff --git a/misc/dumpe2fs.c b/misc/dumpe2fs.c
> index 384ce925..183e2f6f 100644
> --- a/misc/dumpe2fs.c
> +++ b/misc/dumpe2fs.c
> @@ -667,21 +667,8 @@ int main (int argc, char ** argv)
> if (image_dump)
> flags |= EXT2_FLAG_IMAGE_FILE;
> try_open_again:
> - if (use_superblock && !use_blocksize) {
> - for (use_blocksize = EXT2_MIN_BLOCK_SIZE;
> - use_blocksize <= EXT2_MAX_BLOCK_SIZE;
> - use_blocksize *= 2) {
> - retval = ext2fs_open (device_name, flags,
> - use_superblock,
> - use_blocksize, unix_io_manager,
> - &fs);
> - if (!retval)
> - break;
> - }
> - } else {
> - retval = ext2fs_open(device_name, flags, use_superblock,
> - use_blocksize, unix_io_manager, &fs);
> - }
> + retval = ext2fs_open2(device_name, 0, flags, use_superblock,
> + use_blocksize, unix_io_manager, &fs);
> flags |= EXT2_FLAG_IGNORE_CSUM_ERRORS;
> if (retval && !retval_csum) {
> retval_csum = retval;
> --
> 2.14.3
>


Cheers, Andreas






Attachments:
signature.asc (873.00 B)
Message signed with OpenPGP

2019-01-30 03:07:02

by Andreas Dilger

[permalink] [raw]
Subject: Re: [PATCH v7 2/4] e2image: add -b and -B options to use supperblock backup


> On Jan 29, 2019, at 10:51 AM, Artem Blagodarenko <[email protected]> wrote:
>
> e2image has no ability to use superblock backup to copy metadata.
> This feature can be useful if someone wants to make partition
> image and fix it using e2fsck utility.
>
> New -b option allows to pass superblock number, like e2fsck utility does.
> e2image doesn't change primary superblock and store it as is, so
> it can be fixed using e2fsck latter. Option -B allows setting
> superblock size.
>
> Signed-off-by: Artem Blagodarenko <[email protected]>

Reviewed-by: Andreas Dilger <[email protected]>

> ---
> misc/e2image.8.in | 33 +++++++++++++++++++++++++++++++++
> misc/e2image.c | 43 +++++++++++++++++++++++++++++++++++++------
> 2 files changed, 70 insertions(+), 6 deletions(-)
>
> diff --git a/misc/e2image.8.in b/misc/e2image.8.in
> index a7bfdf24..bbbb57ae 100644
> --- a/misc/e2image.8.in
> +++ b/misc/e2image.8.in
> @@ -12,6 +12,15 @@ e2image \- Save critical ext2/ext3/ext4 filesystem metadata to a file
> ]
> [
> .B \-f
> +.B \-b
> +.I superblock
> +]
> +[
> +.B \-B
> +.I blocksize
> +]
> +[
> +.B \-fr
> ]
> .I device
> .I image-file
> @@ -167,6 +176,22 @@ the
> option will prevent analysis of problems related to hash-tree indexed
> directories.
> .PP
> +Option
> +.B \-b
> +.I superblock
> +can be used to get image from partition with broken primary superblock.
> +The partition is copied as-is including broken primary superblock.
> +.PP
> +Option
> +.B \-B
> +.I blocksize
> +can be used to set superblock block size. Normally, e2fsck will search
> +for the superblock at various different block sizes in an attempt to find
> +the appropriate blocksize. This search can be fooled in some cases. This
> +option forces e2fsck to only try locating the superblock at a particular
> +blocksize. If the superblock is not found, e2fsck will terminate with a
> +fatal error.
> +.PP
> Note that this will work even if you substitute "/dev/hda1" for another raw
> disk image, or QCOW2 image previously created by
> .BR e2image .
> @@ -217,6 +242,14 @@ This can be useful to write a qcow2 image containing all data to a
> sparse image file where it can be loop mounted, or to a disk partition.
> Note that this may not work with qcow2 images not generated by e2image.
> .PP
> +Options
> +.B \-b
> +.I superblock
> +and
> +.B \-B
> +.I blocksize
> +can be used same way as for raw images.
> +.PP
> .SH INCLUDING DATA
> Normally
> .B e2image
> diff --git a/misc/e2image.c b/misc/e2image.c
> index 9e21d0db..3c881fee 100644
> --- a/misc/e2image.c
> +++ b/misc/e2image.c
> @@ -104,7 +104,8 @@ static int get_bits_from_size(size_t size)
>
> static void usage(void)
> {
> - fprintf(stderr, _("Usage: %s [ -r|Q ] [ -f ] device image-file\n"),
> + fprintf(stderr, _("Usage: %s [ -r|Q ] [ -f ] [ -b superblock ] [ -B blocksize]"
> + "[ -fr ] device image-file\n"),
> program_name);
> fprintf(stderr, _(" %s -I device image-file\n"), program_name);
> fprintf(stderr, _(" %s -ra [ -cfnp ] [ -o src_offset ] "
> @@ -1267,7 +1268,8 @@ static void output_qcow2_meta_data_blocks(ext2_filsys fs, int fd)
> free_qcow2_image(img);
> }
>
> -static void write_raw_image_file(ext2_filsys fs, int fd, int type, int flags)
> +static void write_raw_image_file(ext2_filsys fs, int fd, int type, int flags,
> + blk64_t superblock)
> {
> struct process_block_struct pb;
> struct ext2_inode inode;
> @@ -1295,6 +1297,22 @@ static void write_raw_image_file(ext2_filsys fs, int fd, int type, int flags)
> }
> }
>
> + if (superblock) {
> + int j;
> +
> + ext2fs_mark_block_bitmap2(meta_block_map, superblock);
> + meta_blocks_count++;
> +
> + /*
> + * Mark the backup superblock descriptors
> + */
> + for (j = 0; j < fs->desc_blocks; j++) {
> + ext2fs_mark_block_bitmap2(meta_block_map,
> + ext2fs_descriptor_block_loc2(fs, superblock, j));
> + }
> + meta_blocks_count += fs->desc_blocks;
> + }
> +
> mark_table_blocks(fs);
> if (show_progress)
> fprintf(stderr, "%s", _("Scanning inodes...\n"));
> @@ -1474,6 +1492,8 @@ int main (int argc, char ** argv)
> int ignore_rw_mount = 0;
> int check = 0;
> struct stat st;
> + blk64_t superblock = 0;
> + int blocksize = 0;
>
> #ifdef ENABLE_NLS
> setlocale(LC_MESSAGES, "");
> @@ -1487,8 +1507,14 @@ int main (int argc, char ** argv)
> if (argc && *argv)
> program_name = *argv;
> add_error_table(&et_ext2_error_table);
> - while ((c = getopt(argc, argv, "nrsIQafo:O:pc")) != EOF)
> + while ((c = getopt(argc, argv, "b:B:nrsIQafo:O:pc")) != EOF)
> switch (c) {
> + case 'b':
> + superblock = strtoull(optarg, NULL, 0);
> + break;
> + case 'B':
> + blocksize = strtoul(optarg, NULL, 0);
> + break;
> case 'I':
> flags |= E2IMAGE_INSTALL_FLAG;
> break;
> @@ -1540,6 +1566,11 @@ int main (int argc, char ** argv)
> "with raw or QCOW2 images."));
> exit(1);
> }
> + if (superblock && !img_type) {
> + com_err(program_name, 0, "%s", _("-b option can only be used "
> + "with raw or QCOW2 images."));
> + exit(1);
> + }
> if ((source_offset || dest_offset) && img_type != E2IMAGE_RAW) {
> com_err(program_name, 0, "%s",
> _("Offsets are only allowed with raw images."));
> @@ -1590,8 +1621,8 @@ int main (int argc, char ** argv)
> }
> }
> sprintf(offset_opt, "offset=%llu", source_offset);
> - retval = ext2fs_open2(device_name, offset_opt, open_flag, 0, 0,
> - unix_io_manager, &fs);
> + retval = ext2fs_open2(device_name, offset_opt, open_flag,
> + superblock, blocksize, unix_io_manager, &fs);
> if (retval) {
> com_err (program_name, retval, _("while trying to open %s"),
> device_name);
> @@ -1681,7 +1712,7 @@ skip_device:
> exit(1);
> }
> if (img_type)
> - write_raw_image_file(fs, fd, img_type, flags);
> + write_raw_image_file(fs, fd, img_type, flags, superblock);
> else
> write_image_file(fs, fd);
>
> --
> 2.14.3
>


Cheers, Andreas






Attachments:
signature.asc (873.00 B)
Message signed with OpenPGP

2019-01-30 03:10:07

by Andreas Dilger

[permalink] [raw]
Subject: Re: [PATCH v7 4/4] ext2fs: automaticlly open backup superblocks

On Jan 29, 2019, at 10:51 AM, Artem Blagodarenko <[email protected]> wrote:
>
> e2image and e2fsck automatically try to open some backup superblocks,
> if only blocksize is set or passed superblock can't be opened.
> Try few backup superblocks (e.g. {1, 3, 5, 7, 9} * blocksize * 8).
>
> This code is moved to lib/support/.
>
> Signed-off-by: Artem Blagodarenko <[email protected]>

Reviewed-by: Andreas Dilger <[email protected]>

> ---
> e2fsck/e2fsck.h | 2 -
> e2fsck/message.c | 3 +-
> e2fsck/unix.c | 14 +++--
> e2fsck/util.c | 73 -------------------------
> lib/support/Makefile.in | 8 ++-
> lib/support/sb_backup.c | 140 ++++++++++++++++++++++++++++++++++++++++++++++++
> lib/support/sb_backup.h | 20 +++++++
> misc/e2image.c | 7 +++
> 8 files changed, 185 insertions(+), 82 deletions(-)
>
> diff --git a/e2fsck/e2fsck.h b/e2fsck/e2fsck.h
> index 1c7a67cb..8576ef24 100644
> --- a/e2fsck/e2fsck.h
> +++ b/e2fsck/e2fsck.h
> @@ -633,8 +633,6 @@ extern void e2fsck_write_inode_full(e2fsck_t ctx, unsigned long ino,
> #ifdef MTRACE
> extern void mtrace_print(char *mesg);
> #endif
> -extern blk64_t get_backup_sb(e2fsck_t ctx, ext2_filsys fs,
> - const char *name, io_manager manager);
> extern int ext2_file_type(unsigned int mode);
> extern int write_all(int fd, char *buf, size_t count);
> void dump_mmp_msg(struct mmp_struct *mmp, const char *fmt, ...)
> diff --git a/e2fsck/message.c b/e2fsck/message.c
> index 727f71d5..ffb40730 100644
> --- a/e2fsck/message.c
> +++ b/e2fsck/message.c
> @@ -465,7 +465,8 @@ static _INLINE_ void expand_percent_expression(FILE *f, ext2_filsys fs,
> fprintf(f, "%*lld", width, (long long) ctx->blkcount);
> break;
> case 'S':
> - fprintf(f, "%llu", get_backup_sb(NULL, fs, NULL, NULL));
> + fprintf(f, "%llu", get_first_backup_sb(NULL, NULL, fs,
> + NULL, NULL));
> break;
> case 's':
> fprintf(f, "%*s", width, ctx->str ? ctx->str : "NULL");
> diff --git a/e2fsck/unix.c b/e2fsck/unix.c
> index ddcf52a4..4ffc039e 100644
> --- a/e2fsck/unix.c
> +++ b/e2fsck/unix.c
> @@ -1489,11 +1489,17 @@ restart:
> retval ? _("Superblock invalid,") :
> _("Group descriptors look bad..."));
> orig_superblock = ctx->superblock;
> - get_backup_sb(ctx, fs, ctx->filesystem_name, io_ptr);
> - if (fs)
> - ext2fs_close_free(&fs);
> orig_retval = retval;
> - retval = try_open_fs(ctx, flags, io_ptr, &fs);
> + retval = try_backups(ctx->filesystem_name,
> + ctx->io_options,
> + flags, &ctx->superblock,
> + &ctx->blocksize, io_ptr, &fs);
> + if (retval == 0) {
> + fs->priv_data = ctx;
> + e2fsck_set_bitmap_type(fs,
> + EXT2FS_BMAP64_RBTREE,
> + "default", NULL);
> + }
> if ((orig_retval == 0) && retval != 0) {
> if (fs)
> ext2fs_close_free(&fs);
> diff --git a/e2fsck/util.c b/e2fsck/util.c
> index db6a1cc1..e2caf3a2 100644
> --- a/e2fsck/util.c
> +++ b/e2fsck/util.c
> @@ -549,79 +549,6 @@ void mtrace_print(char *mesg)
> }
> #endif
>
> -blk64_t get_backup_sb(e2fsck_t ctx, ext2_filsys fs, const char *name,
> - io_manager manager)
> -{
> - struct ext2_super_block *sb;
> - io_channel io = NULL;
> - void *buf = NULL;
> - int blocksize;
> - blk64_t superblock, ret_sb = 8193;
> -
> - if (fs && fs->super) {
> - ret_sb = (fs->super->s_blocks_per_group +
> - fs->super->s_first_data_block);
> - if (ctx) {
> - ctx->superblock = ret_sb;
> - ctx->blocksize = fs->blocksize;
> - }
> - return ret_sb;
> - }
> -
> - if (ctx) {
> - if (ctx->blocksize) {
> - ret_sb = ctx->blocksize * 8;
> - if (ctx->blocksize == 1024)
> - ret_sb++;
> - ctx->superblock = ret_sb;
> - return ret_sb;
> - }
> - ctx->superblock = ret_sb;
> - ctx->blocksize = 1024;
> - }
> -
> - if (!name || !manager)
> - goto cleanup;
> -
> - if (manager->open(name, 0, &io) != 0)
> - goto cleanup;
> -
> - if (ext2fs_get_mem(SUPERBLOCK_SIZE, &buf))
> - goto cleanup;
> - sb = (struct ext2_super_block *) buf;
> -
> - for (blocksize = EXT2_MIN_BLOCK_SIZE;
> - blocksize <= EXT2_MAX_BLOCK_SIZE ; blocksize *= 2) {
> - superblock = blocksize*8;
> - if (blocksize == 1024)
> - superblock++;
> - io_channel_set_blksize(io, blocksize);
> - if (io_channel_read_blk64(io, superblock,
> - -SUPERBLOCK_SIZE, buf))
> - continue;
> -#ifdef WORDS_BIGENDIAN
> - if (sb->s_magic == ext2fs_swab16(EXT2_SUPER_MAGIC))
> - ext2fs_swap_super(sb);
> -#endif
> - if ((sb->s_magic == EXT2_SUPER_MAGIC) &&
> - (EXT2_BLOCK_SIZE(sb) == blocksize)) {
> - ret_sb = superblock;
> - if (ctx) {
> - ctx->superblock = superblock;
> - ctx->blocksize = blocksize;
> - }
> - break;
> - }
> - }
> -
> -cleanup:
> - if (io)
> - io_channel_close(io);
> - if (buf)
> - ext2fs_free_mem(&buf);
> - return (ret_sb);
> -}
> -
> /*
> * Given a mode, return the ext2 file type
> */
> diff --git a/lib/support/Makefile.in b/lib/support/Makefile.in
> index 1d278642..03fdf555 100644
> --- a/lib/support/Makefile.in
> +++ b/lib/support/Makefile.in
> @@ -22,7 +22,8 @@ OBJS= cstring.o \
> quotaio.o \
> quotaio_v2.o \
> quotaio_tree.o \
> - dict.o
> + dict.o \
> + sb_backup.o
>
> SRCS= $(srcdir)/argv_parse.c \
> $(srcdir)/cstring.c \
> @@ -35,7 +36,8 @@ SRCS= $(srcdir)/argv_parse.c \
> $(srcdir)/quotaio.c \
> $(srcdir)/quotaio_tree.c \
> $(srcdir)/quotaio_v2.c \
> - $(srcdir)/dict.c
> + $(srcdir)/dict.c \
> + $(srcdir)/sb_backup.c
>
> LIBRARY= libsupport
> LIBDIR= support
> @@ -168,3 +170,5 @@ quotaio_v2.o: $(srcdir)/quotaio_v2.c $(top_builddir)/lib/config.h \
> $(srcdir)/quotaio_tree.h
> dict.o: $(srcdir)/dict.c $(top_builddir)/lib/config.h \
> $(top_builddir)/lib/dirpaths.h $(srcdir)/dict.h
> +sb_backup.o: $(srcdir)/sb_backup.c $(top_builddir)/lib/config.h \
> + $(srcdir)/sb_backup.h
> diff --git a/lib/support/sb_backup.c b/lib/support/sb_backup.c
> new file mode 100644
> index 00000000..cfa671e5
> --- /dev/null
> +++ b/lib/support/sb_backup.c
> @@ -0,0 +1,140 @@
> +/*
> + * sb_backup.c -- helper functions for getting backup superblocks
> + *
> + * %Begin-Header%
> + * This file may be redistributed under the terms of the GNU Library
> + * General Public License, version 2.
> + * %End-Header%
> + */
> +
> +
> +#include "config.h"
> +#include <stdio.h>
> +#include <string.h>
> +
> +#include "ext2fs/ext2_fs.h"
> +#include "ext2fs/ext2fs.h"
> +
> +blk64_t get_first_backup_sb(blk64_t *superblock, unsigned int *block_size,
> + ext2_filsys fs, const char *name,
> + io_manager manager)
> +{
> + struct ext2_super_block *sb;
> + io_channel io = NULL;
> + void *buf = NULL;
> + int try_blocksize;
> + blk64_t try_superblock, ret_sb = 8193;
> +
> + /* superblock and block_size can be NULL if fs->super is passed */
> + if (fs && fs->super) {
> + ret_sb = fs->super->s_blocks_per_group +
> + fs->super->s_first_data_block;
> + if (superblock)
> + *superblock = ret_sb;
> + if (block_size)
> + *block_size = fs->blocksize;
> + return ret_sb;
> + }
> +
> + if (*block_size) {
> + ret_sb = *block_size * 8;
> + if (*block_size == 1024)
> + ret_sb++;
> + *superblock = ret_sb;
> + return ret_sb;
> + }
> +
> + *superblock = ret_sb;
> + *block_size = 1024;
> +
> + if (!name || !manager)
> + goto cleanup;
> +
> + if (manager->open(name, 0, &io) != 0)
> + goto cleanup;
> +
> + if (ext2fs_get_mem(SUPERBLOCK_SIZE, &buf))
> + goto cleanup;
> + sb = (struct ext2_super_block *) buf;
> +
> + for (try_blocksize = EXT2_MIN_BLOCK_SIZE;
> + try_blocksize <= EXT2_MAX_BLOCK_SIZE ; try_blocksize *= 2) {
> + try_superblock = try_blocksize*8;
> + if (try_blocksize == 1024)
> + try_superblock++;
> + io_channel_set_blksize(io, try_blocksize);
> + if (io_channel_read_blk64(io, try_superblock,
> + -SUPERBLOCK_SIZE, buf))
> + continue;
> +#ifdef WORDS_BIGENDIAN
> + if (sb->s_magic == ext2fs_swab16(EXT2_SUPER_MAGIC))
> + ext2fs_swap_super(sb);
> +#endif
> + if ((sb->s_magic == EXT2_SUPER_MAGIC) &&
> + (EXT2_BLOCK_SIZE(sb) == try_blocksize)) {
> + ret_sb = try_superblock;
> + *superblock = try_superblock;
> + *block_size = try_blocksize;
> + break;
> + }
> + }
> +
> +cleanup:
> + if (io)
> + io_channel_close(io);
> + if (buf)
> + ext2fs_free_mem(&buf);
> + return ret_sb;
> +}
> +
> +errcode_t try_backups(const char *name, const char *io_options,
> + int flags, blk64_t *superblock,
> + unsigned int *block_size, io_manager manager,
> + ext2_filsys *ret_fs)
> +{
> + errcode_t retval, retval2;
> + blk64_t try_block_number;
> + unsigned int i;
> +
> + /*
> + * Get first superblock location based on heuristic.
> + * Blocksize is also returned and used to find next
> + * superblock copy location.
> + */
> + try_block_number = get_first_backup_sb(superblock, block_size,
> + *ret_fs, name, manager);
> + retval = ext2fs_open2(name, io_options, flags, try_block_number,
> + *block_size, manager, ret_fs);
> + if (!retval)
> + return 0;
> +
> + /*
> + * Try 3d, 5th, 7th, 9th superblock copy
> + */
> + for (i = 3; i <= 9; i += 2) {
> + try_block_number = (i * (*block_size) * 8);
> + if (*block_size == 1024)
> + try_block_number++;
> + if (*ret_fs) {
> + ext2fs_free(*ret_fs);
> + *ret_fs = NULL;
> + }
> + retval2 = ext2fs_open2(name, io_options, flags,
> + try_block_number, *block_size, manager,
> + ret_fs);
> + /*
> + * Partition is too small to have this superblock copy.
> + * Return previous reading return code
> + */
> + if (retval2 == EXT2_ET_SHORT_READ)
> + break;
> +
> + retval = retval2;
> + if (!retval) {
> + *superblock = try_block_number;
> + break;
> + }
> + }
> +
> + return retval;
> +}
> diff --git a/lib/support/sb_backup.h b/lib/support/sb_backup.h
> new file mode 100644
> index 00000000..1e18e37c
> --- /dev/null
> +++ b/lib/support/sb_backup.h
> @@ -0,0 +1,20 @@
> +/*
> + * sb_backup.h -- helper functions for getting backup superblocks
> + *
> + * %Begin-Header%
> + * This file may be redistributed under the terms of the GNU Library
> + * General Public License, version 2.
> + * %End-Header%
> + */
> +
> +#include "ext2fs/ext2_fs.h"
> +#include "ext2fs/ext2fs.h"
> +
> +blk64_t get_first_backup_sb(blk64_t *superblock, unsigned int *block_size,
> + ext2_filsys fs, const char *name,
> + io_manager manager);
> +
> +errcode_t try_backups(const char *name, const char *io_options,
> + int flags, blk64_t *superblock,
> + unsigned int *block_size, io_manager manager,
> + ext2_filsys *ret_fs);
> diff --git a/misc/e2image.c b/misc/e2image.c
> index 3c881fee..786282ec 100644
> --- a/misc/e2image.c
> +++ b/misc/e2image.c
> @@ -1623,6 +1623,13 @@ int main (int argc, char ** argv)
> sprintf(offset_opt, "offset=%llu", source_offset);
> retval = ext2fs_open2(device_name, offset_opt, open_flag,
> superblock, blocksize, unix_io_manager, &fs);
> + if (retval & (superblock | blocksize)) {
> + printf(_("Try backups in other location.\n"));
> + retval = try_backups(device_name, offset_opt, open_flag,
> + &superblock, &blocksize,
> + unix_io_manager, &fs);
> + printf(_("Use superblock %i.\n"), superblock);
> + }
> if (retval) {
> com_err (program_name, retval, _("while trying to open %s"),
> device_name);
> --
> 2.14.3
>


Cheers, Andreas






Attachments:
signature.asc (873.00 B)
Message signed with OpenPGP

2019-03-06 16:48:57

by Theodore Ts'o

[permalink] [raw]
Subject: Re: [PATCH v7 1/4] ext2fs: opening filesystem code refactoring

On Tue, Jan 29, 2019 at 08:51:31PM +0300, Artem Blagodarenko wrote:
> There are similar opening filesystem code in different utilities.
>
> The patch moves improved handling from try_open_fs()
> into ext2fs_open(). This function make one of the action
> based on parameters:
> 1) open filesystem with given superblock, superblock size
> 2) open filesystem with given superblock, but try to
> find right block size

This changes the behavior of ext2fs_open() when the superblock and
blocksize parameters is zero. Previously we read the primary
superblock at the known location, and a blocksize of zero means "don't
care" and so we just do the right thing. With your patch, it's going
to loop for all possible block sizes, failing with
EXT2_ET_UNEXPECTED_BLOCK_SIZE, until we stumble on the correct file
system. That's unfortunate, especially if the device is opened with
IO_FLAG_DIRECT_IO.

So the brute force searching should be done using if an explicit flag
is given. Using a new flag to request this new behavior is safer, but
it's probably OK if we only do the blocksize search if the superblock
parameter is zero --- you'll note that's what the code you were
replacing in dumpe2fs and e2fsck was doing.

More generally, for this whole patch series, in the "happy case" where
the file system is fine, e2fsck and e2image should *not* do any extra
work.

- Ted

2019-03-06 17:06:47

by Theodore Ts'o

[permalink] [raw]
Subject: Re: [PATCH v7 2/4] e2image: add -b and -B options to use supperblock backup

On Tue, Jan 29, 2019 at 08:51:32PM +0300, Artem Blagodarenko wrote:
> e2image has no ability to use superblock backup to copy metadata.
> This feature can be useful if someone wants to make partition
> image and fix it using e2fsck utility.
>
> New -b option allows to pass superblock number, like e2fsck utility does.
> e2image doesn't change primary superblock and store it as is, so
> it can be fixed using e2fsck latter. Option -B allows setting
> superblock size.
>
> Signed-off-by: Artem Blagodarenko <[email protected]>

This change is independent of the first patch, so I've applied it,
thanks.

- Ted

2019-03-06 17:07:01

by Theodore Ts'o

[permalink] [raw]
Subject: Re: [PATCH v7 1/4] ext2fs: opening filesystem code refactoring

On Tue, Jan 29, 2019 at 08:51:31PM +0300, Artem Blagodarenko wrote:
> There are similar opening filesystem code in different utilities.
>
> The patch moves improved handling from try_open_fs()
> into ext2fs_open(). This function make one of the action
> based on parameters:
> 1) open filesystem with given superblock, superblock size
> 2) open filesystem with given superblock, but try to
> find right block size
>
> Signed-off-by: Artem Blagodarenko <[email protected]>

Thanks, applied.

- Ted