search for: nr_thread

...<errno.h> #include <getopt.h> #include <pthread.h> +#include <linux/falloc.h> #include <lzma.h> #define DEBUG 0 @@ -145,10 +146,11 @@ usage (int exitcode) static void xzfile_uncompress (const char *filename, const char *outputfile, unsigned nr_threads) { int fd, ofd; + off_t hole_start, data_start; uint64_t size; lzma_index *idx; /* Open the file. */ fd = open (filename, O_RDONLY); @@ -176,10 +178,29 @@ xzfile_uncompress (const char *filename, const char *outputfile, posix_fadvise (fd, 0, 0, POSIX_FADV_RANDOM|POSIX_FADV_DON...

These two patches make appliance building thread-safe. The first adds a test which easily demonstrates the problem. The second fixes the issue. For more information see Ian McLeod's analysis of the bug here: https://bugzilla.redhat.com/show_bug.cgi?id=790528#c5 Rich.

I'm not going to put this upstream because there's no benefit. However it is useful to record the patch on the mailing list. Rich.

--- src/inspect-apps.c | 13 ++++++++++++- src/inspect-fs-windows.c | 6 ++++++ src/journal.c | 5 +++++ 3 files changed, 23 insertions(+), 1 deletion(-) diff --git a/src/inspect-apps.c b/src/inspect-apps.c index 20cf00a..8fbae9c 100644 --- a/src/inspect-apps.c +++ b/src/inspect-apps.c @@ -35,11 +35,22 @@ #include <sys/endian.h> #endif -/* be32toh is usually a macro

The way we used posix_fadvise was wrong, and yet right! Rich.

...++- 7 files changed, 74 insertions(+), 27 deletions(-) diff --git a/align/scan.c b/align/scan.c index 5f1d98a..e7327ea 100644 --- a/align/scan.c +++ b/align/scan.c @@ -87,6 +87,7 @@ usage (int status) " --help Display brief help\n" " -P nr_threads Use at most nr_threads\n" " -q|--quiet No output, just exit code\n" + " --uuid Print UUIDs instead of names\n" " -v|--verbose Verbose messages\n" " -V|--vers...

--- builder/pxzcat-c.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/builder/pxzcat-c.c b/builder/pxzcat-c.c index 0bbd296..dec9cc2 100644 --- a/builder/pxzcat-c.c +++ b/builder/pxzcat-c.c @@ -214,8 +214,10 @@ pxzcat (value filenamev, value outputfilev, unsigned nr_threads) unix_error (err, (char *) "ftruncate", outputfilev); } +#if defined HAVE_POSIX_FADVISE /* Tell the kernel we won't read the output file. */ ignore_value (posix_fadvise (fd, 0, 0, POSIX_FADV_RANDOM|POSIX_FADV_DONTNEED)); +#endif /* Iterate over blocks. */ iter...

Various changes/fixes to use smaller stack frames. Rich.

...+++++++++++++++++++++++++++++++++++++++ 4 files changed, 95 insertions(+), 19 deletions(-) diff --git a/builder/pxzcat-c.c b/builder/pxzcat-c.c index 1f5ceeb..44722bc 100644 --- a/builder/pxzcat-c.c +++ b/builder/pxzcat-c.c @@ -214,10 +214,7 @@ pxzcat (value filenamev, value outputfilev, unsigned nr_threads) unix_error (err, (char *) "ftruncate", outputfilev); } -#if defined HAVE_POSIX_FADVISE - /* Tell the kernel we won't read the output file. */ - ignore_value (posix_fadvise (fd, 0, 0, POSIX_FADV_RANDOM|POSIX_FADV_DONTNEED)); -#endif + guestfs_int_fadvise_noreuse (fd);...

More posix_fadvise stuff, and document what Linux really does with these calls. Also fixes a nasty bug in virt-builder. Rich.

...trm.avail_in > BUFSIZ) strm.avail_in = BUFSIZ; - n = read (fd, &buf, strm.avail_in); + n = read (fd, buf, strm.avail_in); if (n == -1) unix_error (errno, (char *) "read", filenamev); @@ -454,12 +459,17 @@ iter_blocks (lzma_index *idx, unsigned nr_threads, value filenamev, int fd, value outputfilev, int ofd) { struct global_state global; - struct per_thread_state per_thread[nr_threads]; - pthread_t thread[nr_threads]; + CLEANUP_FREE struct per_thread_state *per_thread = NULL; + CLEANUP_FREE pthread_t *thread = NULL; unsigne...

From: Nikolay Ivanets <stenavin@gmail.com> This patch depends on changes in 'common' sub-module posted here: https://www.redhat.com/archives/libguestfs/2020-February/msg00096.html Nikolay Ivanets (1): tools: add '--blocksize' option for C-based tools align/scan.c | 8 ++++++++ align/virt-alignment-scan.pod | 12 ++++++++++++ cat/cat.c

This is very useful for debugging multithreaded programs. Rich.

...at[=raw|..] Force disk format for -a option\n" + " --blocksize[=512|4096]\n" + " Set sector size of the disk for -a option\n" " --help Display brief help\n" " -P nr_threads Use at most nr_threads\n" " -q|--quiet No output, just exit code\n" @@ -119,6 +121,7 @@ main (int argc, char *argv[]) { "connect", 1, 0, 'c' }, { "domain", 1, 0, 'd' }, { "format", 2, 0, 0...

From: Nikolay Ivanets <stenavin@gmail.com> This patch depends on changes in 'common' sub-module posted here: https://www.redhat.com/archives/libguestfs/2020-February/msg00099.html v2: Almost the same as v1 except '--blocksize' option description is moved into a common submodule (similar to key-option.pod). v1 was here:

...at[=raw|..] Force disk format for -a option\n" + " --blocksize[=512|4096]\n" + " Set sector size of the disk for -a option\n" " --help Display brief help\n" " -P nr_threads Use at most nr_threads\n" " -q|--quiet No output, just exit code\n" @@ -119,6 +121,7 @@ main (int argc, char *argv[]) { "connect", 1, 0, 'c' }, { "domain", 1, 0, 'd' }, { "format", 2, 0, 0...

...Specify libvirt URI for -d option\n" - " -d|--domain guest Add disks from libvirt guest\n" - " --format[=raw|..] Force disk format for -a option\n" - " --help Display brief help\n" - " -P nr_threads Use at most nr_threads\n" - " -q|--quiet No output, just exit code\n" - " -v|--verbose Verbose messages\n" - " -V|--version Display version and exit\n" - " -x...

Hi, this series update libguestfs to a recent gnulib version, so that we can use its new getprogname module, and solve altogether one of the porting issues (the need for 'program_name' by the error module of gnulib), and have a single way to get the name of the current program. A number of changes in tools mostly, although mechanical. Thanks, Pino Toscano (3): Update gnulib to latest

Since the first patch: - The program is now called 'guestunmount'. - I tested the --fd option and it appears to work. - You can now control retries / quiet. - Revised man pages. - Includes tests. I'm just running through the automated tests now. Rich.

This version doesn't crash or cause hung processes or stuck mountpoints, so that's an improvement. Rich.