[PATCH] Reduce GUESTFS_PRIVATE usage
by Pino Toscano
Remove the GUESTFS_PRIVATE=1 define for some tools and tests which don't
really use any private API.
---
align/Makefile.am | 1 -
df/Makefile.am | 1 -
tests/protocol/Makefile.am | 1 -
tests/qemu/Makefile.am | 2 --
tests/regressions/Makefile.am | 6 ++----
5 files changed, 2 insertions(+), 9 deletions(-)
diff --git a/align/Makefile.am b/align/Makefile.am
index 04eea16..833739a 100644
--- a/align/Makefile.am
+++ b/align/Makefile.am
@@ -50,7 +50,6 @@ virt_alignment_scan_SOURCES = \
virt_alignment_scan_CPPFLAGS = \
-DGUESTFS_WARN_DEPRECATED=1 \
- -DGUESTFS_PRIVATE=1 \
-I$(top_srcdir)/src -I$(top_builddir)/src \
-I$(top_srcdir)/df \
-I$(top_srcdir)/fish \
diff --git a/df/Makefile.am b/df/Makefile.am
index 256263d..83879cd 100644
--- a/df/Makefile.am
+++ b/df/Makefile.am
@@ -54,7 +54,6 @@ virt_df_SOURCES = \
virt_df_CPPFLAGS = \
-DGUESTFS_WARN_DEPRECATED=1 \
- -DGUESTFS_PRIVATE=1 \
-DLOCALEBASEDIR=\""$(datadir)/locale"\" \
-I$(top_srcdir)/src -I$(top_builddir)/src \
-I$(top_srcdir)/fish \
diff --git a/tests/protocol/Makefile.am b/tests/protocol/Makefile.am
index 8494fa0..4cdd96f 100644
--- a/tests/protocol/Makefile.am
+++ b/tests/protocol/Makefile.am
@@ -43,7 +43,6 @@ check_PROGRAMS = test-error-messages
test_error_messages_SOURCES = \
test-error-messages.c
test_error_messages_CPPFLAGS = \
- -DGUESTFS_PRIVATE=1 \
-I$(top_srcdir)/gnulib/lib -I$(top_builddir)/gnulib/lib \
-I$(top_srcdir)/src -I$(top_builddir)/src
test_error_messages_CFLAGS = \
diff --git a/tests/qemu/Makefile.am b/tests/qemu/Makefile.am
index 091361f..fe45d88 100644
--- a/tests/qemu/Makefile.am
+++ b/tests/qemu/Makefile.am
@@ -43,7 +43,6 @@ qemu_boot_SOURCES = \
../../df/estimate-max-threads.h \
qemu-boot.c
qemu_boot_CPPFLAGS = \
- -DGUESTFS_PRIVATE=1 \
-I$(top_srcdir)/gnulib/lib -I$(top_builddir)/gnulib/lib \
-I$(top_srcdir)/src -I$(top_builddir)/src \
-I$(top_srcdir)/df
@@ -61,7 +60,6 @@ qemu_boot_LDADD = \
qemu_speed_test_SOURCES = \
qemu-speed-test.c
qemu_speed_test_CPPFLAGS = \
- -DGUESTFS_PRIVATE=1 \
-I$(top_srcdir)/gnulib/lib -I$(top_builddir)/gnulib/lib \
-I$(top_srcdir)/src -I$(top_builddir)/src \
-I$(top_srcdir)/df
diff --git a/tests/regressions/Makefile.am b/tests/regressions/Makefile.am
index 3aae57b..ec97553 100644
--- a/tests/regressions/Makefile.am
+++ b/tests/regressions/Makefile.am
@@ -130,8 +130,7 @@ rhbz914931_LDADD = \
rhbz1055452_SOURCES = rhbz1055452.c
rhbz1055452_CPPFLAGS = \
- -I$(top_srcdir)/src -I$(top_builddir)/src \
- -DGUESTFS_PRIVATE=1
+ -I$(top_srcdir)/src -I$(top_builddir)/src
rhbz1055452_CFLAGS = \
-pthread \
$(WARN_CFLAGS) $(WERROR_CFLAGS)
@@ -140,8 +139,7 @@ rhbz1055452_LDADD = \
test_big_heap_SOURCES = test-big-heap.c
test_big_heap_CPPFLAGS = \
- -I$(top_srcdir)/src -I$(top_builddir)/src \
- -DGUESTFS_PRIVATE=1
+ -I$(top_srcdir)/src -I$(top_builddir)/src
test_big_heap_CFLAGS = \
$(WARN_CFLAGS) $(WERROR_CFLAGS)
test_big_heap_LDADD = \
--
2.5.5
8 years, 9 months
[PATCH supermin 0/5] Make supermin mini-initrd quieter and faster.
by Richard W.M. Jones
Various patches to make supermin quieter. By outputting fewer
messages on the fast path, we use the slow emulated UART less, and
this improves boot times.
Also remove some kernel modules that we cannot or should not be using,
which also improves boot times.
Rich.
8 years, 9 months
[PATCH] ffind API to retrieve a file name given its inode
by Matteo Cafasso
The ffind API allows to retrieve a file name from a device given its inode.
The function returns a struct "tsknode" which contains the file name, its inode and it's allocation status. The struct will be employed by other APIs as well (fls, ifind etc..).
$ ./run guestfish --ro -a /home/noxdafox/disks/ubuntu.qcow2
><fs> run
><fs> ffind /dev/sda1 2
tsk_name: /
tsk_inode: 2
tsk_allocated: 1
><fs> ffind /dev/sda1 3
tsk_name: /usr/bin/
tsk_inode: 3
tsk_allocated: 0
/usr/bin/ has been reallocated to node 786577
><fs> mount /dev/sda1 /
><fs> stat /usr/bin/
dev: 2049
ino: 786577
...
Matteo Cafasso (1):
added ffind API
daemon/tsk.c | 60 ++++++++++++++++++++++++++++++++++++++++++++++++++++
generator/actions.ml | 18 ++++++++++++++++
generator/structs.ml | 14 ++++++++++--
src/MAX_PROC_NR | 2 +-
4 files changed, 91 insertions(+), 3 deletions(-)
--
2.7.0
8 years, 9 months
[PATCH] conn: Pretend to be a serial terminal, so sgabios doesn't hang.
by Richard W.M. Jones
This tedious workaround avoids a 0.26 second pause when using sgabios
(the Serial Graphics Adapter). It's basically a workaround for buggy
code in sgabios, but much easier than fixing the assembler.
---
src/conn-socket.c | 32 +++++++++++++++++++++++++++++++-
1 file changed, 31 insertions(+), 1 deletion(-)
diff --git a/src/conn-socket.c b/src/conn-socket.c
index 5b6b80e..14da260 100644
--- a/src/conn-socket.c
+++ b/src/conn-socket.c
@@ -33,6 +33,8 @@
#include <assert.h>
#include <libintl.h>
+#include "ignore-value.h"
+
#include "guestfs.h"
#include "guestfs-internal.h"
@@ -314,6 +316,9 @@ handle_log_message (guestfs_h *g,
{
CLEANUP_FREE char *buf = safe_malloc (g, BUFSIZ);
ssize_t n;
+ const char dsr_request[] = "\033[6n";
+ const char dsr_reply[] = "\033[24;80R";
+ const char dsr_reply_padding[] = "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b";
/* Carried over from ancient proto.c code. The comment there was:
*
@@ -341,7 +346,32 @@ handle_log_message (guestfs_h *g,
return -1;
}
- /* It's an actual log message, send it upwards. */
+ /* It's an actual log message. */
+
+ /* SGABIOS tries to query the "serial console" for its size using the
+ * ISO/IEC 6429 Device Status Report (ESC [ 6 n). If it doesn't
+ * read anything back, then it unfortunately hangs for 0.25 seconds.
+ * Therefore we detect this situation and send back a fake console
+ * size.
+ */
+ if (memmem (buf, n, dsr_request, sizeof dsr_request - 1) != NULL) {
+ /* Ignore any error from this write, as it's just an optimization.
+ * We can't even be sure that console_sock is a socket or that
+ * it's writable.
+ */
+ ignore_value (write (conn->console_sock, dsr_reply,
+ sizeof dsr_reply - 1));
+ /* Additionally, because of a bug in sgabios, it will still pause
+ * unless you write at least 14 bytes, so we have to pad the
+ * reply. We can't pad with NULs since sgabios's input routine
+ * ignores these, so we have to use some other safe padding
+ * characters. Backspace seems innocuous.
+ */
+ ignore_value (write (conn->console_sock, dsr_reply_padding,
+ sizeof dsr_reply_padding - 1));
+ }
+
+ /* Send it upwards. */
guestfs_int_log_message_callback (g, buf, n);
return 1;
--
2.7.4
8 years, 9 months
[PATCH 0/7] Small portability changes
by Pino Toscano
Assorted collection of small improvements in making libguestfs build on
non-Linux OSes; most of the changes impact tests though.
Thanks,
Pino Toscano (7):
build: check the path of fuser, and use it in FUSE code
tests: move guestfs-md5.sh to test-data
v2v: tests: isolate SHA1 calculation in an own shared function
v2v: tests: use guestfs-hashsums.sh for MD5
php: pass $(MAKE) to run-php-tests.sh
build: check the path of true, and use it in tests
build: improve GUESTFS_FIND_DB_TOOL macro
fuse/guestunmount.c | 6 ++---
fuse/test-fuse.c | 2 +-
m4/guestfs-find-db-tool.m4 | 7 +++---
m4/guestfs_progs.m4 | 8 +++++++
php/Makefile.am | 4 +++-
php/run-php-tests.sh | 2 +-
test-data/Makefile.am | 3 +++
test-data/guestfs-hashsums.sh | 43 +++++++++++++++++++++++++++++++++++
tests/qemu/Makefile.am | 1 -
tests/qemu/guestfs-md5.sh | 30 ------------------------
tests/qemu/qemu-liveness.sh | 2 +-
tests/qemu/qemu-snapshot-isolation.sh | 2 +-
tests/regressions/rhbz790721.c | 2 +-
v2v/test-v2v-i-ova-formats.sh | 4 +++-
v2v/test-v2v-i-ova-gz.sh | 4 +++-
v2v/test-v2v-i-ova-two-disks.sh | 6 +++--
v2v/test-v2v-i-ova.sh | 6 +++--
v2v/test-v2v-in-place.sh | 6 +++--
18 files changed, 87 insertions(+), 51 deletions(-)
create mode 100755 test-data/guestfs-hashsums.sh
delete mode 100755 tests/qemu/guestfs-md5.sh
--
2.5.0
8 years, 9 months
[PATCH] tests/qemu: Add program for tracing and analyzing boot times.
by Richard W.M. Jones
---
.gitignore | 1 +
tests/qemu/Makefile.am | 22 +-
tests/qemu/boot-analysis.c | 1098 ++++++++++++++++++++++++++++++++++++++++++++
3 files changed, 1118 insertions(+), 3 deletions(-)
create mode 100644 tests/qemu/boot-analysis.c
diff --git a/.gitignore b/.gitignore
index 1493011..9306a19 100644
--- a/.gitignore
+++ b/.gitignore
@@ -510,6 +510,7 @@ Makefile.in
/tests/mountable/test-internal-parse-mountable
/tests/parallel/test-parallel
/tests/protocol/test-error-messages
+/tests/qemu/boot-analysis
/tests/qemu/qemu-boot
/tests/qemu/qemu-speed-test
/tests/regressions/rhbz501893
diff --git a/tests/qemu/Makefile.am b/tests/qemu/Makefile.am
index 6c72b18..57ceeae 100644
--- a/tests/qemu/Makefile.am
+++ b/tests/qemu/Makefile.am
@@ -34,10 +34,10 @@ EXTRA_DIST = \
qemu-boot.c \
qemu-speed-test.c
-# qemu-boot & qemu-speed-test are built but not run by default as they
-# are mainly qemu & kernel diagnostic tools.
+# qemu-boot, qemu-speed-test and boot-analysis are built but not run
+# by default as they are mainly qemu & kernel diagnostic tools.
-check_PROGRAMS = qemu-boot qemu-speed-test
+check_PROGRAMS = qemu-boot qemu-speed-test boot-analysis
qemu_boot_SOURCES = \
../../df/estimate-max-threads.c \
@@ -76,6 +76,22 @@ qemu_speed_test_LDADD = \
$(LTLIBINTL) \
$(top_builddir)/gnulib/lib/libgnu.la
+boot_analysis_SOURCES = \
+ boot-analysis.c
+boot_analysis_CPPFLAGS = \
+ -I$(top_srcdir)/gnulib/lib -I$(top_builddir)/gnulib/lib \
+ -I$(top_srcdir)/src -I$(top_builddir)/src
+boot_analysis_CFLAGS = \
+ $(WARN_CFLAGS) $(WERROR_CFLAGS)
+boot_analysis_LDADD = \
+ $(top_builddir)/src/libutils.la \
+ $(top_builddir)/src/libguestfs.la \
+ $(LIBXML2_LIBS) \
+ $(LIBVIRT_LIBS) \
+ $(LTLIBINTL) \
+ $(top_builddir)/gnulib/lib/libgnu.la \
+ -lm
+
# Don't run these tests in parallel, since they are designed to check
# the integrity of qemu.
.NOTPARALLEL:
diff --git a/tests/qemu/boot-analysis.c b/tests/qemu/boot-analysis.c
new file mode 100644
index 0000000..4bc9e28
--- /dev/null
+++ b/tests/qemu/boot-analysis.c
@@ -0,0 +1,1098 @@
+/* libguestfs
+ * Copyright (C) 2016 Red Hat Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+ */
+
+/* Trace and analyze the appliance boot process to find out which
+ * steps are taking the most time. It is not part of the standard
+ * tests.
+ *
+ * This needs to be run on a quiet machine, so that other processes
+ * disturb the timing as little as possible. The program is
+ * completely safe to run at any time. It doesn't read or write any
+ * external files, and it doesn't require root.
+ *
+ * You can run it from the build directory like this:
+ *
+ * make
+ * make -C tests/qemu boot-analysis
+ * ./run tests/qemu/boot-analysis
+ *
+ * The way it works is roughly like this:
+ *
+ * We create a libguestfs handle and register callback handlers so we
+ * can see appliance messages, trace events and so on.
+ *
+ * We then launch the handle and shut it down as quickly as possible.
+ *
+ * While the handle is running, events (seen by the callback handlers)
+ * are written verbatim into an in-memory buffer, with timestamps.
+ *
+ * Afterwards we analyze the result using regular expressions to try
+ * to identify a "timeline" for the handle (eg. at what time did the
+ * BIOS hand control to the kernel).
+ *
+ * The whole process is repeated across a few runs, and the final
+ * timeline (including statistical analysis of the variation between
+ * runs) gets printed.
+ *
+ * The program is very sensitive to the specific messages printed by
+ * BIOS/kernel/supermin/userspace, so it won't work on non-x86, and it
+ * will require periodic adjustment of the regular expressions in
+ * order to keep things up to date.
+ */
+
+#include <config.h>
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <inttypes.h>
+#include <string.h>
+#include <unistd.h>
+#include <getopt.h>
+#include <limits.h>
+#include <time.h>
+#include <errno.h>
+#include <error.h>
+#include <ctype.h>
+#include <assert.h>
+#include <math.h>
+
+#include "ignore-value.h"
+
+#include "guestfs.h"
+#include "guestfs-internal-frontend.h"
+
+#define NR_WARMUP_PASSES 3
+#define NR_TEST_PASSES 5
+#define DEBUG 0
+
+/* Per-pass data collected. */
+struct pass_data {
+ size_t pass;
+ struct timespec start_t;
+ struct timespec end_t;
+ int64_t elapsed_ns;
+
+ /* Array of timestamped events. */
+ size_t nr_events;
+ struct event *events;
+
+ /* Was the previous appliance log message incomplete? If so, this
+ * contains the index of that incomplete message in the events
+ * array.
+ */
+ ssize_t incomplete_log_message;
+
+ /* Have we seen the launch event yet? We don't record events until
+ * this one has been received. This makes it easy to base the
+ * timeline at event 0.
+ */
+ int seen_launch;
+};
+
+struct event {
+ struct timespec t;
+ uint64_t source;
+ char *message;
+};
+
+static struct pass_data pass_data[NR_TEST_PASSES];
+
+/* The final timeline consisting of various activities starting and
+ * ending. We're interested in when the activities start, and how
+ * long they take (mean, variance, standard deviation of length).
+ */
+struct activity {
+ char *name; /* Name of this activity. */
+ int flags;
+#define LONG_ACTIVITY 1 /* Expected to take a long time. */
+
+ /* For each pass, record the actual start & end events of this
+ * activity.
+ */
+ size_t start_event[NR_TEST_PASSES];
+ size_t end_event[NR_TEST_PASSES];
+
+ double t; /* Start (ns offset). */
+
+ /* Length of this activity. */
+ double mean; /* Mean time elapsed (ns). */
+ double variance; /* Variance. */
+ double sd; /* Standard deviation. */
+ double percent; /* Percent of total elapsed time. */
+
+ double end_t; /* t + mean - 1 */
+};
+
+static size_t nr_activities;
+static struct activity *activities;
+
+static void run_test (void);
+static void get_time (struct timespec *ts);
+static int64_t timespec_diff (const struct timespec *x, const struct timespec *y);
+static struct event *add_event (struct pass_data *, uint64_t source);
+static guestfs_h *create_handle (void);
+static void set_up_event_handlers (guestfs_h *g, size_t pass);
+static void add_drive (guestfs_h *g);
+static void check_pass_data (void);
+static void dump_pass_data (void);
+static void construct_timeline (void);
+static void dump_timeline (void);
+static void print_info (void);
+static void print_analysis (void);
+static void free_pass_data (void);
+static void free_final_timeline (void);
+static void ansi_green (void);
+static void ansi_red (void);
+static void ansi_blue (void);
+static void ansi_magenta (void);
+static void ansi_restore (void);
+
+static void
+usage (int exitcode)
+{
+ fprintf (stderr,
+ "boot-analysis: Trace and analyze the appliance boot process.\n"
+ "Usage:\n"
+ " boot-analysis [--options]\n"
+ "Options:\n"
+ " --help Display this usage text and exit.\n");
+ exit (exitcode);
+}
+
+int
+main (int argc, char *argv[])
+{
+ enum { HELP_OPTION = CHAR_MAX + 1 };
+ static const char *options = "";
+ static const struct option long_options[] = {
+ { "help", 0, 0, HELP_OPTION },
+ { 0, 0, 0, 0 }
+ };
+ int c, option_index;
+
+ for (;;) {
+ c = getopt_long (argc, argv, options, long_options, &option_index);
+ if (c == -1) break;
+
+ switch (c) {
+ case 0:
+ /* Options which are long only. */
+ fprintf (stderr, "%s: unknown long option: %s (%d)\n",
+ guestfs_int_program_name, long_options[option_index].name, option_index);
+ exit (EXIT_FAILURE);
+
+ case HELP_OPTION:
+ usage (EXIT_SUCCESS);
+
+ default:
+ usage (EXIT_FAILURE);
+ }
+ }
+
+ if (STRNEQ (host_cpu, "x86_64"))
+ fprintf (stderr, "WARNING: host_cpu != x86_64: This program may not work or give bogus results.\n");
+
+ run_test ();
+}
+
+static void
+run_test (void)
+{
+ guestfs_h *g;
+ size_t i;
+
+ printf ("Warming up the libguestfs cache ...\n");
+ for (i = 0; i < NR_WARMUP_PASSES; ++i) {
+ g = create_handle ();
+ add_drive (g);
+ if (guestfs_launch (g) == -1)
+ exit (EXIT_FAILURE);
+ guestfs_close (g);
+ }
+
+ printf ("Running the tests in %d passes ...\n", NR_TEST_PASSES);
+ for (i = 0; i < NR_TEST_PASSES; ++i) {
+ g = create_handle ();
+ set_up_event_handlers (g, i);
+ add_drive (g);
+ if (guestfs_launch (g) == -1)
+ exit (EXIT_FAILURE);
+ guestfs_close (g);
+
+ printf (" pass %zu: %zu events collected in %" PRIi64 " ns\n",
+ i+1, pass_data[i].nr_events, pass_data[i].elapsed_ns);
+ }
+
+ if (DEBUG)
+ dump_pass_data ();
+
+ printf ("Analyzing the results ...\n");
+ check_pass_data ();
+ construct_timeline ();
+
+ if (DEBUG)
+ dump_timeline ();
+
+ printf ("\n");
+ print_info ();
+ printf ("\n");
+ print_analysis ();
+
+ free_pass_data ();
+ free_final_timeline ();
+}
+
+static void
+get_time (struct timespec *ts)
+{
+ if (clock_gettime (CLOCK_REALTIME, ts) == -1)
+ error (EXIT_FAILURE, errno, "clock_gettime: CLOCK_REALTIME");
+}
+
+/* Computes Y - X, returning nanoseconds. */
+static int64_t
+timespec_diff (const struct timespec *x, const struct timespec *y)
+{
+ int64_t nsec;
+
+ nsec = (y->tv_sec - x->tv_sec) * UINT64_C(1000000000);
+ nsec += y->tv_nsec - x->tv_nsec;
+ return nsec;
+}
+
+static struct event *
+add_event (struct pass_data *data, uint64_t source)
+{
+ struct event *ret;
+
+ data->nr_events++;
+ data->events = realloc (data->events,
+ sizeof (struct event) * data->nr_events);
+ if (data->events == NULL)
+ error (EXIT_FAILURE, errno, "realloc");
+ ret = &data->events[data->nr_events-1];
+ get_time (&ret->t);
+ ret->source = source;
+ ret->message = NULL;
+ return ret;
+}
+
+/* Common function to create the handle and set various defaults. */
+static guestfs_h *
+create_handle (void)
+{
+ guestfs_h *g;
+
+ g = guestfs_create ();
+ if (!g) error (EXIT_FAILURE, errno, "guestfs_create");
+
+ /* We always run these tests using LIBGUESTFS_BACKEND=direct. It
+ * may be in future we need to test libvirt as well, in case
+ * performance issues are suspected there, but so far libvirt has
+ * not been a bottleneck.
+ */
+ if (guestfs_set_backend (g, "direct") == -1)
+ exit (EXIT_FAILURE);
+
+ return g;
+}
+
+/* Common function to add the /dev/null drive. */
+static void
+add_drive (guestfs_h *g)
+{
+ if (guestfs_add_drive (g, "/dev/null") == -1)
+ exit (EXIT_FAILURE);
+}
+
+/* Called when the handle is closed. Perform any cleanups required in
+ * the pass_data here.
+ */
+static void
+close_callback (guestfs_h *g, void *datavp, uint64_t source,
+ int eh, int flags,
+ const char *buf, size_t buf_len,
+ const uint64_t *array, size_t array_len)
+{
+ struct pass_data *data = datavp;
+ struct event *event;
+
+ if (!data->seen_launch)
+ return;
+
+ event = add_event (data, source);
+ event->message = strdup ("close callback");
+ if (event->message == NULL)
+ error (EXIT_FAILURE, errno, "strdup");
+
+ get_time (&data->end_t);
+ data->elapsed_ns = timespec_diff (&data->start_t, &data->end_t);
+}
+
+/* Called when the qemu subprocess exits.
+ * XXX This is never called - why?
+ */
+static void
+subprocess_quit_callback (guestfs_h *g, void *datavp, uint64_t source,
+ int eh, int flags,
+ const char *buf, size_t buf_len,
+ const uint64_t *array, size_t array_len)
+{
+ struct pass_data *data = datavp;
+ struct event *event;
+
+ if (!data->seen_launch)
+ return;
+
+ event = add_event (data, source);
+ event->message = strdup ("subprocess quit callback");
+ if (event->message == NULL)
+ error (EXIT_FAILURE, errno, "strdup");
+}
+
+/* Called when the launch operation is complete (the library and the
+ * guestfs daemon and talking to each other).
+ */
+static void
+launch_done_callback (guestfs_h *g, void *datavp, uint64_t source,
+ int eh, int flags,
+ const char *buf, size_t buf_len,
+ const uint64_t *array, size_t array_len)
+{
+ struct pass_data *data = datavp;
+ struct event *event;
+
+ if (!data->seen_launch)
+ return;
+
+ event = add_event (data, source);
+ event->message = strdup ("launch done callback");
+ if (event->message == NULL)
+ error (EXIT_FAILURE, errno, "strdup");
+}
+
+/* Called when we get (possibly part of) a log message (or more than
+ * one log message) from the appliance (which may include qemu, the
+ * BIOS, kernel, etc).
+ */
+static void
+appliance_callback (guestfs_h *g, void *datavp, uint64_t source,
+ int eh, int flags,
+ const char *buf, size_t buf_len,
+ const uint64_t *array, size_t array_len)
+{
+ struct pass_data *data = datavp;
+ struct event *event;
+ size_t i, len, slen;
+
+ if (!data->seen_launch)
+ return;
+
+ /* If the previous log message was incomplete then we may need to
+ * append part of the current log message to a previous one.
+ */
+ if (data->incomplete_log_message >= 0) {
+ len = buf_len;
+ for (i = 0; i < buf_len; ++i) {
+ if (buf[i] == '\n') {
+ len = i;
+ break;
+ }
+ }
+
+ event = &data->events[data->incomplete_log_message];
+ slen = strlen (event->message);
+ event->message = realloc (event->message, slen + len + 1);
+ if (event->message == NULL)
+ error (EXIT_FAILURE, errno, "realloc");
+ memcpy (event->message + slen, buf, len);
+ event->message[slen + len] = '\0';
+
+ /* Skip what we just added to the previous incomplete message. */
+ buf += len;
+ buf_len -= len;
+
+ if (buf_len == 0) /* still not complete, more to come! */
+ return;
+
+ /* Skip the \n in the buffer. */
+ buf++;
+ buf_len--;
+ data->incomplete_log_message = -1;
+ }
+
+ /* Add the event, or perhaps multiple events if the message
+ * contains \n characters.
+ */
+ while (buf_len > 0) {
+ len = buf_len;
+ for (i = 0; i < buf_len; ++i) {
+ if (buf[i] == '\n') {
+ len = i;
+ break;
+ }
+ }
+
+ event = add_event (data, source);
+ event->message = strndup (buf, len);
+ if (event->message == NULL)
+ error (EXIT_FAILURE, errno, "strndup");
+
+ /* Skip what we just added to the event. */
+ buf += len;
+ buf_len -= len;
+
+ if (buf_len == 0) {
+ /* Event is incomplete (doesn't end with \n). We'll finish it
+ * in the next callback.
+ */
+ data->incomplete_log_message = event - data->events;
+ return;
+ }
+
+ /* Skip the \n in the buffer. */
+ buf++;
+ buf_len--;
+ }
+}
+
+/* Called when we get a debug message from the library side. These
+ * are always delivered as complete messages.
+ */
+static void
+library_callback (guestfs_h *g, void *datavp, uint64_t source,
+ int eh, int flags,
+ const char *buf, size_t buf_len,
+ const uint64_t *array, size_t array_len)
+{
+ struct pass_data *data = datavp;
+ struct event *event;
+
+ if (!data->seen_launch)
+ return;
+
+ event = add_event (data, source);
+ event->message = strndup (buf, buf_len);
+ if (event->message == NULL)
+ error (EXIT_FAILURE, errno, "strndup");
+}
+
+/* Called when we get a call trace message (a libguestfs API function
+ * has been called or is returning). These are always delivered as
+ * complete messages.
+ */
+static void
+trace_callback (guestfs_h *g, void *datavp, uint64_t source,
+ int eh, int flags,
+ const char *buf, size_t buf_len,
+ const uint64_t *array, size_t array_len)
+{
+ struct pass_data *data = datavp;
+ struct event *event;
+ char *message;
+
+ message = strndup (buf, buf_len);
+ if (message == NULL)
+ error (EXIT_FAILURE, errno, "strndup");
+
+ if (STREQ (message, "launch"))
+ data->seen_launch = 1;
+
+ if (!data->seen_launch) {
+ free (message);
+ return;
+ }
+
+ event = add_event (data, source);
+ event->message = message;
+}
+
+/* Common function to set up event callbacks and record data in memory
+ * for a particular pass (0 <= pass < NR_TEST_PASSES).
+ */
+static void
+set_up_event_handlers (guestfs_h *g, size_t pass)
+{
+ struct pass_data *data;
+
+ assert (/* 0 <= pass && */ pass < NR_TEST_PASSES);
+
+ data = &pass_data[pass];
+ data->pass = pass;
+ data->nr_events = 0;
+ data->events = NULL;
+ get_time (&data->start_t);
+ data->incomplete_log_message = -1;
+ data->seen_launch = 0;
+
+ guestfs_set_event_callback (g, close_callback,
+ GUESTFS_EVENT_CLOSE, 0, data);
+ guestfs_set_event_callback (g, subprocess_quit_callback,
+ GUESTFS_EVENT_SUBPROCESS_QUIT, 0, data);
+ guestfs_set_event_callback (g, launch_done_callback,
+ GUESTFS_EVENT_LAUNCH_DONE, 0, data);
+ guestfs_set_event_callback (g, appliance_callback,
+ GUESTFS_EVENT_APPLIANCE, 0, data);
+ guestfs_set_event_callback (g, library_callback,
+ GUESTFS_EVENT_LIBRARY, 0, data);
+ guestfs_set_event_callback (g, trace_callback,
+ GUESTFS_EVENT_TRACE, 0, data);
+
+ guestfs_set_verbose (g, 1);
+ guestfs_set_trace (g, 1);
+}
+
+/* Sanity check the collected events. */
+static void
+check_pass_data (void)
+{
+ size_t i, j;
+ int64_t ns;
+ const char *message;
+
+ for (i = 0; i < NR_TEST_PASSES; ++i) {
+ assert (pass_data[i].pass == i);
+ assert (pass_data[i].elapsed_ns > 1000);
+ assert (pass_data[i].nr_events > 0);
+ assert (pass_data[i].events != NULL);
+
+ for (j = 0; j < pass_data[i].nr_events; ++j) {
+ assert (pass_data[i].events[j].t.tv_sec > 0);
+ if (j > 0) {
+ ns = timespec_diff (&pass_data[i].events[j-1].t,
+ &pass_data[i].events[j].t);
+ assert (ns >= 0);
+ }
+ assert (pass_data[i].events[j].source != 0);
+ message = pass_data[i].events[j].message;
+ assert (message != NULL);
+ assert (strlen (message) > 0);
+ assert (strchr (message, '\n') == NULL);
+ }
+ }
+}
+
+static void
+print_escaped_string (const char *message)
+{
+ while (*message) {
+ if (isprint (*message))
+ putchar (*message);
+ else
+ printf ("\\x%02x", (unsigned int) *message);
+ message++;
+ }
+}
+
+/* Dump the events to stdout, if DEBUG is set. */
+static void
+dump_pass_data (void)
+{
+ size_t i, j;
+
+ for (i = 0; i < NR_TEST_PASSES; ++i) {
+ printf ("pass %zu\n", pass_data[i].pass);
+ printf (" number of events collected %zu\n", pass_data[i].nr_events);
+ printf (" elapsed time %" PRIi64 " ns\n", pass_data[i].elapsed_ns);
+ for (j = 0; j < pass_data[i].nr_events; ++j) {
+ int64_t ns;
+ CLEANUP_FREE char *event_str = NULL;
+
+ ns = timespec_diff (&pass_data[i].start_t, &pass_data[i].events[j].t);
+ event_str = guestfs_event_to_string (pass_data[i].events[j].source);
+ printf (" #%zu: +%" PRIi64 " [%s] \"", j, ns, event_str);
+ print_escaped_string (pass_data[i].events[j].message);
+ printf ("\"\n");
+ }
+ }
+}
+
+/* Add an activity to the global list. */
+static struct activity *
+add_activity (const char *name, int flags)
+{
+ struct activity *ret;
+
+ nr_activities++;
+ activities = realloc (activities, sizeof (struct activity) * nr_activities);
+ if (activities == NULL)
+ error (EXIT_FAILURE, errno, "realloc");
+ ret = &activities[nr_activities-1];
+ ret->name = strdup (name);
+ if (ret->name == NULL)
+ error (EXIT_FAILURE, errno, "strdup");
+ ret->flags = flags;
+ return ret;
+}
+
+static struct activity *
+find_activity (const char *name)
+{
+ size_t i;
+
+ for (i = 0; i < nr_activities; ++i)
+ if (STREQ (activities[i].name, name))
+ return &activities[i];
+ error (EXIT_FAILURE, 0,
+ "internal error: could not find activity '%s'", name);
+ /*NOTREACHED*/
+ abort ();
+}
+
+static int
+compare_activities (const void *av, const void *bv)
+{
+ const struct activity *a = av;
+ const struct activity *b = bv;
+
+ return a->t - b->t;
+}
+
+/* Analyze significant events from the events array, to form a
+ * timeline of activities.
+ */
+static void
+construct_timeline (void)
+{
+ size_t i, j, k;
+ struct pass_data *data;
+ struct activity *activity;
+ int64_t delta_ns;
+
+ for (i = 0; i < NR_TEST_PASSES; ++i) {
+ data = &pass_data[i];
+
+#define FIND(name, flags, begin_cond, end_cond) \
+ do { \
+ activity = NULL; \
+ for (j = 0; j < data->nr_events; ++j) { \
+ if (begin_cond) { \
+ for (k = j+1; k < data->nr_events; ++k) { \
+ if (end_cond) { \
+ if (i == 0) \
+ activity = add_activity (name, flags); \
+ else \
+ activity = find_activity (name); \
+ break; \
+ } \
+ } \
+ break; \
+ } \
+ } \
+ if (activity) { \
+ activity->start_event[i] = j; \
+ activity->end_event[i] = k; \
+ } \
+ else \
+ error (EXIT_FAILURE, 0, "could not find activity '%s' in pass '%zu'", \
+ name, i); \
+ } while (0)
+
+#define FIND_MULTIPLE(debug_name, flags, next_cond, end_cond) \
+ do { \
+ activity = NULL; \
+ for (j = 0; j < data->nr_events; ++j) { \
+ if (next_cond) { \
+ if (activity) \
+ activity->end_event[i] = j; \
+ if (i == 0) \
+ activity = add_activity (data->events[j].message, flags); \
+ else \
+ activity = find_activity (data->events[j].message); \
+ activity->start_event[i] = j; \
+ } \
+ else if (end_cond) \
+ break; \
+ } \
+ if (j < data->nr_events && activity) \
+ activity->end_event[i] = j; \
+ else \
+ error (EXIT_FAILURE, 0, "could not find activity '%s' in pass '%zu'", \
+ debug_name, i); \
+ } while (0)
+
+ /* Add one activity which is going to cover the whole process
+ * from launch to close. The launch event is always event 0.
+ * NB: This activity must be called "run" (see below).
+ */
+ FIND ("run", LONG_ACTIVITY,
+ j == 0, data->events[k].source == GUESTFS_EVENT_CLOSE);
+
+ /* Find where we invoke supermin --build. This should be a null
+ * operation, but it still takes time to run the external command.
+ */
+ FIND ("supermin:build", 0,
+ data->events[j].source == GUESTFS_EVENT_LIBRARY &&
+ strstr (data->events[j].message,
+ "begin building supermin appliance"),
+ data->events[k].source == GUESTFS_EVENT_LIBRARY &&
+ strstr (data->events[k].message,
+ "finished building supermin appliance"));
+
+ /* Find where we invoke qemu to test features. */
+ FIND ("qemu:feature-detect", 0,
+ data->events[j].source == GUESTFS_EVENT_LIBRARY &&
+ strstr (data->events[j].message,
+ "begin testing qemu features"),
+ data->events[k].source == GUESTFS_EVENT_LIBRARY &&
+ strstr (data->events[k].message,
+ "finished testing qemu features"));
+
+ /* Find where we run qemu. */
+ FIND ("qemu", LONG_ACTIVITY,
+ data->events[j].source == GUESTFS_EVENT_APPLIANCE &&
+ strstr (data->events[j].message, "-nodefconfig"),
+ data->events[k].source == GUESTFS_EVENT_CLOSE);
+
+ /* From starting qemu up to entering the BIOS is the qemu overhead. */
+ FIND ("qemu:overhead", 0,
+ data->events[j].source == GUESTFS_EVENT_APPLIANCE &&
+ strstr (data->events[j].message, "-nodefconfig"),
+ data->events[k].source == GUESTFS_EVENT_APPLIANCE &&
+ strstr (data->events[k].message, "SeaBIOS (version"));
+
+ /* From entering the BIOS to starting the kernel is the BIOS overhead. */
+ FIND ("seabios:overhead", 0,
+ data->events[j].source == GUESTFS_EVENT_APPLIANCE &&
+ strstr (data->events[j].message, "SeaBIOS (version"),
+ data->events[k].source == GUESTFS_EVENT_APPLIANCE &&
+ strstr (data->events[k].message, "Linux version"));
+
+ /* Find where we run the guest kernel. */
+ FIND ("kernel", LONG_ACTIVITY,
+ data->events[j].source == GUESTFS_EVENT_APPLIANCE &&
+ strstr (data->events[j].message, "Linux version"),
+ data->events[k].source == GUESTFS_EVENT_CLOSE);
+
+ /* XXX More detailed look at kernel boot stages here. */
+
+ /* Find where we run supermin mini-initrd. */
+ FIND ("supermin:mini-initrd", 0,
+ data->events[j].source == GUESTFS_EVENT_APPLIANCE &&
+ strstr (data->events[j].message, "supermin:") &&
+ strstr (data->events[j].message, "starting up"),
+ data->events[k].source == GUESTFS_EVENT_APPLIANCE &&
+ strstr (data->events[k].message, "supermin: chroot"));
+
+ /* Loading kernel modules from supermin initrd. */
+ FIND_MULTIPLE
+ ("supermin insmod", 0,
+ data->events[j].source == GUESTFS_EVENT_APPLIANCE &&
+ strstr (data->events[j].message, "supermin: internal insmod"),
+ data->events[j].source == GUESTFS_EVENT_APPLIANCE &&
+ strstr (data->events[j].message, "supermin: picked"));
+
+ /* Find where we run the /init script. */
+ FIND ("/init", 0,
+ data->events[j].source == GUESTFS_EVENT_APPLIANCE &&
+ strstr (data->events[j].message, "supermin: chroot"),
+ data->events[k].source == GUESTFS_EVENT_APPLIANCE &&
+ strstr (data->events[k].message, "guestfsd --verbose"));
+
+ /* Everything from the chroot to the first echo in the /init
+ * script counts as bash overhead.
+ */
+ FIND ("bash:overhead", 0,
+ data->events[j].source == GUESTFS_EVENT_APPLIANCE &&
+ strstr (data->events[j].message, "supermin: chroot"),
+ data->events[k].source == GUESTFS_EVENT_APPLIANCE &&
+ strstr (data->events[k].message, "Starting /init script"));
+
+ /* XXX More detail from /init script. */
+
+ /* Find where we run guestfsd. */
+ FIND ("guestfsd", 0,
+ data->events[j].source == GUESTFS_EVENT_APPLIANCE &&
+ strstr (data->events[j].message, "guestfsd --verbose"),
+ data->events[k].source == GUESTFS_EVENT_APPLIANCE &&
+ strstr (data->events[k].message, "fsync /dev/sda"));
+
+ /* Shutdown process. */
+ FIND ("shutdown", 0,
+ data->events[j].source == GUESTFS_EVENT_TRACE &&
+ STREQ (data->events[j].message, "close"),
+ data->events[k].source == GUESTFS_EVENT_CLOSE);
+ }
+
+ /* Now we can go through the activities, computing the start and
+ * elapsed time.
+ */
+ for (j = 0; j < nr_activities; ++j) {
+ activity = &activities[j];
+
+ activity->t = 0;
+ activity->mean = 0;
+ for (i = 0; i < NR_TEST_PASSES; ++i) {
+ delta_ns =
+ timespec_diff (&pass_data[i].events[0].t,
+ &pass_data[i].events[activity->start_event[i]].t);
+ activity->t += delta_ns;
+
+ delta_ns =
+ timespec_diff (&pass_data[i].events[activity->start_event[i]].t,
+ &pass_data[i].events[activity->end_event[i]].t);
+ activity->mean += delta_ns;
+ }
+
+ /* Divide through to get real start time and mean of each activity. */
+ activity->t /= NR_TEST_PASSES;
+ activity->mean /= NR_TEST_PASSES;
+
+ /* Calculate the end time of this activity. It's convenient when
+ * drawing the timeline for one activity to finish just before the
+ * next activity starts, rather than having them end and start at
+ * the same time, hence ``- 1'' here.
+ */
+ activity->end_t = activity->t + activity->mean - 1;
+
+ /* The above only calculated mean. Now we are able to
+ * calculate from the mean the variance and the standard
+ * deviation.
+ */
+ activity->variance = 0;
+ for (i = 0; i < NR_TEST_PASSES; ++i) {
+ delta_ns =
+ timespec_diff (&pass_data[i].events[activity->start_event[i]].t,
+ &pass_data[i].events[activity->end_event[i]].t);
+ activity->variance += pow (delta_ns - activity->mean, 2);
+ }
+ activity->variance /= NR_TEST_PASSES;
+
+ activity->sd = sqrt (activity->variance);
+ }
+
+ /* Get the total mean elapsed time from the special "run" activity. */
+ activity = find_activity ("run");
+ for (j = 0; j < nr_activities; ++j)
+ activities[j].percent = 100.0 * activities[j].mean / activity->mean;
+
+ /* Sort the activities by start time. */
+ qsort (activities, nr_activities, sizeof (struct activity),
+ compare_activities);
+}
+
+/* Dump the timeline to stdout, if DEBUG is set. */
+static void
+dump_timeline (void)
+{
+ size_t i;
+
+ for (i = 0; i < nr_activities; ++i) {
+ printf ("activity %zu:\n", i);
+ printf (" name = %s\n", activities[i].name);
+ printf (" start - end = %.1f - %.1f\n",
+ activities[i].t, activities[i].end_t);
+ printf (" mean elapsed = %.1f\n", activities[i].mean);
+ printf (" variance = %.1f\n", activities[i].variance);
+ printf (" s.d = %.1f\n", activities[i].sd);
+ printf (" percent = %.1f\n", activities[i].percent);
+ }
+}
+
+/* Print some information that will allow us to determine the test
+ * system when reviewing the results in future.
+ */
+static void
+print_info (void)
+{
+ size_t i;
+
+ printf ("%s %s\n", PACKAGE_NAME, PACKAGE_VERSION_FULL);
+
+ printf ("Host:\n");
+ ignore_value (system ("uname -a"));
+ ignore_value (system ("grep '^model name' /proc/cpuinfo | head -1"));
+
+ /* We can dig some information about qemu and the appliance out of
+ * the events.
+ */
+ printf ("Appliance:\n");
+ assert (NR_TEST_PASSES > 0);
+ for (i = 0; i < pass_data[0].nr_events; ++i) {
+ const char *message = pass_data[0].events[i].message;
+ if (strstr (message, "qemu version") ||
+ (strstr (message, "SeaBIOS") && strstr (message, "version")) ||
+ strstr (message, "Linux version") ||
+ (strstr (message, "supermin") && strstr (message, "starting up"))) {
+ print_escaped_string (message);
+ putchar ('\n');
+ }
+ }
+}
+
+static void
+print_analysis (void)
+{
+ double t = -1; /* Current time. */
+ /* Which columns contain activities that we are displaying now?
+ * -1 == unused column, else index of an activity
+ */
+ CLEANUP_FREE ssize_t *columns = NULL;
+ const size_t nr_columns = nr_activities;
+ size_t last_free_column = 0;
+
+ size_t i, j;
+ double smallest_next_t;
+ const double MAX_T = 1e20;
+
+ columns = malloc (nr_columns * sizeof (ssize_t));
+ if (columns == NULL) error (EXIT_FAILURE, errno, "malloc");
+ for (j = 0; j < nr_columns; ++j)
+ columns[j] = -1;
+
+ for (;;) {
+ /* Find the next significant time to display, which is a time when
+ * some activity started or ended.
+ */
+ smallest_next_t = MAX_T;
+ for (i = 0; i < nr_activities; ++i) {
+ if (t < activities[i].t && activities[i].t < smallest_next_t)
+ smallest_next_t = activities[i].t;
+ else if (t < activities[i].end_t && activities[i].end_t < smallest_next_t)
+ smallest_next_t = activities[i].end_t;
+ }
+ if (smallest_next_t == MAX_T)
+ break; /* Finished. */
+
+ t = smallest_next_t;
+
+ /* If there are any activities that ended before this time, drop
+ * them from the columns list.
+ */
+ for (i = 0; i < nr_activities; ++i) {
+ if (activities[i].end_t < t) {
+ for (j = 0; j < nr_columns; ++j)
+ if (columns[j] == (ssize_t) i) {
+ columns[j] = -1;
+ break;
+ }
+ }
+ }
+
+ /* May need to adjust last_free_column after previous operation. */
+ while (last_free_column > 0 && columns[last_free_column-1] == -1)
+ last_free_column--;
+
+ /* If there are any activities starting at this time, add them to
+ * the right hand end of the columns list.
+ */
+ for (i = 0; i < nr_activities; ++i) {
+ if (activities[i].t == t)
+ columns[last_free_column++] = i;
+ }
+
+ /* Draw the line. */
+ ansi_blue ();
+ printf ("%1.6fs: ", t / 1000000000);
+
+ ansi_magenta ();
+ for (j = 0; j < last_free_column; ++j) {
+ if (columns[j] >= 0) {
+ if (activities[columns[j]].t == t)
+ printf ("↑ ");
+ else if (activities[columns[j]].end_t == t)
+ printf ("↓ ");
+ else
+ printf ("│ ");
+ }
+ else
+ printf (" ");
+ }
+ ansi_restore ();
+
+ for (j = 0; j < last_free_column; ++j) {
+ if (columns[j] >= 0 && activities[columns[j]].t == t) /* ↑ */ {
+ int warning =
+ !(activities[columns[j]].flags & LONG_ACTIVITY) &&
+ activities[columns[j]].percent >= 5.0;
+
+ if (warning) ansi_red (); else ansi_green ();
+ print_escaped_string (activities[columns[j]].name);
+ ansi_restore ();
+ printf (" mean:%1.6fs ±%.1fms ",
+ activities[columns[j]].mean / 1000000000,
+ activities[columns[j]].sd / 1000000);
+ if (warning) ansi_red (); else ansi_green ();
+ printf ("(%.1f%%) ", activities[columns[j]].percent);
+ ansi_restore ();
+ }
+ }
+
+ printf ("\n");
+ }
+}
+
+/* Free the non-static part of the pass_data structures. */
+static void
+free_pass_data (void)
+{
+ size_t i, j;
+
+ for (i = 0; i < NR_TEST_PASSES; ++i) {
+ for (j = 0; j < pass_data[i].nr_events; ++j)
+ free (pass_data[i].events[j].message);
+ free (pass_data[i].events);
+ }
+}
+
+static void
+free_final_timeline (void)
+{
+ size_t i;
+
+ for (i = 0; i < nr_activities; ++i)
+ free (activities[i].name);
+ free (activities);
+}
+
+/* Colours. */
+static void
+ansi_green (void)
+{
+ if (isatty (1))
+ fputs ("\033[0;32m", stdout);
+}
+
+static void
+ansi_red (void)
+{
+ if (isatty (1))
+ fputs ("\033[1;31m", stdout);
+}
+
+static void
+ansi_blue (void)
+{
+ if (isatty (1))
+ fputs ("\033[1;34m", stdout);
+}
+
+static void
+ansi_magenta (void)
+{
+ if (isatty (1))
+ fputs ("\033[1;35m", stdout);
+}
+
+static void
+ansi_restore (void)
+{
+ if (isatty (1))
+ fputs ("\033[0m", stdout);
+}
--
2.5.0
8 years, 9 months
Fwd: libguestfs patch
by Richard W.M. Jones
----- Forwarded message from "Bernhard M. Wiedemann" <bwiedemann(a)suse.de> -----
Date: Fri, 18 Mar 2016 15:41:16 +0100
From: "Bernhard M. Wiedemann" <bwiedemann(a)suse.de>
To: rjones(a)redhat.com
Subject: libguestfs patch
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101
Thunderbird/38.6.0
Hi,
here is a patch for making libguestfs build more reproducibly
Ciao
Bernhard M.
Index: libguestfs-1.26.10/generator/docstrings.ml
===================================================================
--- libguestfs-1.26.10.orig/generator/docstrings.ml
+++ libguestfs-1.26.10/generator/docstrings.ml
@@ -51,9 +51,7 @@ fact that they are deprecated indicates
with correct use of these functions." prefix alt in
Some txt
-let copyright_years =
- let this_year = 1900 + (localtime (time ())).tm_year in
- if this_year > 2009 then sprintf "2009-%04d" this_year else "2009"
+let copyright_years = "2009-2016"
(* Generate a header block in a number of standard styles. *)
type comment_style =
----- End forwarded message -----
I don't think I'd accept this as-is, but it would be OK if the final
year (ie. 2016) was placed in configure.ac, and passed to the
generator via a config.ml file. This is because it's good to have all
the version / date stuff in one place, and that place is at the top of
configure.ac.
Sadly we don't yet have a generator/config.ml{,.in} file, but one
could be easily created.
Rich.
--
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
virt-p2v converts physical machines to virtual machines. Boot with a
live CD or over the network (PXE) and turn machines into KVM guests.
http://libguestfs.org/virt-v2v
8 years, 9 months
[PATCH supermin] init: Add a blacklist of kmods that we want to exclude from the mini initrd.
by Richard W.M. Jones
We want to exclude virtio-gpu since it's irrelevant, large, and pulls
in other unwanted dependencies (modeswitching, drm). Add a second
list of kmods which is a blacklist, applied after the first.
This reduces the libguestfs initrd size from 39M down to 17M, with
concomitant small reductions in boot time.
---
src/ext2_initrd.ml | 10 +++++++++-
1 file changed, 9 insertions(+), 1 deletion(-)
diff --git a/src/ext2_initrd.ml b/src/ext2_initrd.ml
index 63ed493..a177018 100644
--- a/src/ext2_initrd.ml
+++ b/src/ext2_initrd.ml
@@ -54,6 +54,13 @@ let kmods = [
"megaraid*.ko*";
]
+(* A blacklist of kmods which match the above patterns, but which we
+ * subsequently remove.
+ *)
+let not_kmods = [
+ "virtio-gpu.ko*";
+]
+
let rec build_initrd debug tmpdir modpath initrd =
if debug >= 1 then
printf "supermin: ext2: creating minimal initrd '%s'\n%!" initrd;
@@ -73,7 +80,8 @@ let rec build_initrd debug tmpdir modpath initrd =
fun topset modl ->
let m = Filename.basename modl in
let matches wildcard = fnmatch wildcard m [FNM_PATHNAME] in
- if List.exists matches kmods then
+ if List.exists matches kmods && not (List.exists matches not_kmods)
+ then
StringSet.add modl topset
else
topset
--
2.7.3
8 years, 9 months
[PATCH v3 0/5] v2v: more control over device types
by Roman Kagan
The decision on which device type to use for disks, network and video
cards on output used to be taken deep inside the converting functions.
This is not always desirable. In particular, there are scenarios when
this decision is made before the convertion takes place. E.g. in
in-place mode, the decisions are taken and the output VM configuration
is created outside of v2v tool.
This patchset adds support for such scenarios. Specifically,
- the input configuration parsers are taught to extract network and
video device models in addition to storage controllers
- the converting functions are taught to take the requested device types
into consideration
- the main routine in in-place mode takes the relevant data from the
input VM and passes it to the converter
The copying mode may eventually also profit from this patchset by taking
command-line options for controlling device types, but this isn't
included here.
---
v3:
- assume that source config in --in-place mode can't have more than one
storage, network, and video types
- add patch to better explain --in-place
v2:
- add catch-all string-valued variants for source network and video adapter
models
- use match instead of mixing match and if
Roman Kagan (5):
v2v: collect source network and video adapter types
v2v: introduce requested guestcaps type
v2v: take requested caps into account when converting
v2v: in-place: request caps based on source config
v2v: better explain --in-place
test-data/phony-guests/guests.xml.in | 8 ++++
v2v/convert_linux.ml | 49 ++++++++++++++-----
v2v/convert_windows.ml | 4 +-
v2v/input_disk.ml | 2 +
v2v/input_libvirtxml.ml | 27 +++++++++++
v2v/input_ova.ml | 2 +
v2v/modules_list.ml | 3 +-
v2v/modules_list.mli | 3 +-
v2v/test-v2v-i-ova-formats.expected | 1 +
v2v/test-v2v-i-ova-gz.expected | 1 +
v2v/test-v2v-i-ova-two-disks.expected | 1 +
v2v/test-v2v-print-source.sh | 4 +-
v2v/types.ml | 77 +++++++++++++++++++++++++-----
v2v/types.mli | 21 ++++++++-
v2v/v2v.ml | 66 ++++++++++++++++++++++++--
v2v/virt-v2v.pod | 3 +-
v2v/windows_virtio.ml | 89 ++++++++++++++++++++++++++---------
v2v/windows_virtio.mli | 6 +++
18 files changed, 310 insertions(+), 57 deletions(-)
--
2.5.0
8 years, 9 months
