Avoid overflowing the column-width calc printing files over 4B blocks.
Document the [KMG] suffixes for the "-b <blocksize>" option.
The blocksize is limited to at most 1GiB blocksize to avoid shifting
all extents down to zero GB in size. Even the use of 1GB blocksize
is unlikely, but non-ext4 filesystems may use multi-GB extents.
Signed-off-by: Andreas Dilger <[email protected]>
Lustre-bug-id: https://jira.whamcloud.com/browse/LU-13197
---
misc/filefrag.8.in | 4 ++--
misc/filefrag.c | 36 ++++++++++++++++++++++++------------
2 files changed, 26 insertions(+), 14 deletions(-)
diff --git a/misc/filefrag.8.in b/misc/filefrag.8.in
index 5292672..4b89e72 100644
--- a/misc/filefrag.8.in
+++ b/misc/filefrag.8.in
@@ -33,8 +33,8 @@ testing purposes.
.BI \-b blocksize
Use
.I blocksize
-in bytes for output instead of the filesystem blocksize.
-For compatibility with earlier versions of
+in bytes, or with [KMG] suffix, up to 1GB for output instead of the
+filesystem blocksize. For compatibility with earlier versions of
.BR filefrag ,
if
.I blocksize
diff --git a/misc/filefrag.c b/misc/filefrag.c
index 1eec146..032535f 100644
--- a/misc/filefrag.c
+++ b/misc/filefrag.c
@@ -53,7 +53,7 @@ extern int optind;
#include <ext2fs/fiemap.h>
int verbose = 0;
-int blocksize; /* Use specified blocksize (default 1kB) */
+unsigned int blocksize; /* Use specified blocksize (default 1kB) */
int sync_file = 0; /* fsync file before getting the mapping */
int xattr_map = 0; /* get xattr mapping */
int force_bmap; /* force use of FIBMAP instead of FIEMAP */
@@ -73,7 +73,7 @@ const char *hex_fmt = "%4d: %*llx..%*llx: %*llx..%*llx: %6llx: %s\n";
#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
#define EXT3_IOC_GETFLAGS _IOR('f', 1, long)
-static int int_log2(int arg)
+static int ulong_log2(unsigned long arg)
{
int l = 0;
@@ -85,7 +85,7 @@ static int int_log2(int arg)
return l;
}
-static int int_log10(unsigned long long arg)
+static int ulong_log10(unsigned long long arg)
{
int l = 0;
@@ -452,17 +452,17 @@ static int frag_report(const char *filename)
}
last_device = st.st_dev;
- width = int_log10(fsinfo.f_blocks);
+ width = ulong_log10(fsinfo.f_blocks);
if (width > physical_width)
physical_width = width;
numblocks = (st.st_size + blksize - 1) / blksize;
if (blocksize != 0)
- blk_shift = int_log2(blocksize);
+ blk_shift = ulong_log2(blocksize);
else
- blk_shift = int_log2(blksize);
+ blk_shift = ulong_log2(blksize);
- width = int_log10(numblocks);
+ width = ulong_log10(numblocks);
if (width > logical_width)
logical_width = width;
if (verbose)
@@ -517,7 +517,7 @@ out_close:
static void usage(const char *progname)
{
- fprintf(stderr, "Usage: %s [-b{blocksize}] [-BeksvxX] file ...\n",
+ fprintf(stderr, "Usage: %s [-b{blocksize}[KMG]] [-BeksvxX] file ...\n",
progname);
exit(1);
}
@@ -535,7 +535,9 @@ int main(int argc, char**argv)
case 'b':
if (optarg) {
char *end;
- blocksize = strtoul(optarg, &end, 0);
+ unsigned long val;
+
+ val = strtoul(optarg, &end, 0);
if (end) {
#if __GNUC_PREREQ (7, 0)
#pragma GCC diagnostic push
@@ -544,15 +546,15 @@ int main(int argc, char**argv)
switch (end[0]) {
case 'g':
case 'G':
- blocksize *= 1024;
+ val *= 1024;
/* fall through */
case 'm':
case 'M':
- blocksize *= 1024;
+ val *= 1024;
/* fall through */
case 'k':
case 'K':
- blocksize *= 1024;
+ val *= 1024;
break;
default:
break;
@@ -561,6 +563,16 @@ int main(int argc, char**argv)
#pragma GCC diagnostic pop
#endif
}
+ /* Specifying too large a blocksize will just
+ * shift all extents down to zero length. Even
+ * 1GB is questionable, but caveat emptor. */
+ if (val > 1024 * 1024 * 1024) {
+ fprintf(stderr,
+ "%s: blocksize %lu over 1GB\n",
+ argv[0], val);
+ usage(argv[0]);
+ }
+ blocksize = val;
} else { /* Allow -b without argument for compat. Remove
* this eventually so "-b {blocksize}" works */
fprintf(stderr, "%s: -b needs a blocksize "
--
1.8.0
On Feb 11, 2020, at 5:58 PM, Andreas Dilger <[email protected]> wrote:
>
> Avoid overflowing the column-width calc printing files over 4B blocks.
>
> Document the [KMG] suffixes for the "-b <blocksize>" option.
>
> The blocksize is limited to at most 1GiB blocksize to avoid shifting
> all extents down to zero GB in size. Even the use of 1GB blocksize
> is unlikely, but non-ext4 filesystems may use multi-GB extents.
>
> Signed-off-by: Andreas Dilger <[email protected]>
> Lustre-bug-id: https://jira.whamcloud.com/browse/LU-13197
Sorry,
this one was a repeat. I actually meant to actually send the other patch
"e2fsck: avoid overflow with very large dirs", which was accidentally
dropped from my previous patch series.
Cheers, Andreas