Patches 2 upstream systemd #32430
Patches 2 upstream systemd #32430
Conversation
rajmohanrgithub
force-pushed
the
patches-2-upstream-systemd
branch
2 times, most recently
from
6f51013
to
d866b87
Compare
* unit.c: log unit names and timestamp of unit state change * manager.c: log time stamp when boot is completed * main.c: log time stamp with systemd internal states Signed-off-by: rajmohan r <rajmohan.r@kpit.com>
+ Scale the x-axis of the resulting plot by a factor (default 1.0) + Add activation timestamps to each bar Signed-off-by: rajmohan r <rajmohan.r@kpit.com>
We want core dumps from regularly skipped files (changing credentials or lacking read permission) Signed-off-by: rajmohan r <rajmohan.r@kpit.com>
rajmohanrgithub
force-pushed
the
patches-2-upstream-systemd
branch
from
afd48d6
to
9e5a453
Compare
If a mount point is not a directory, systemd will remove() it and then mkdir_p() the full path. Signed-off-by: rajmohan r <rajmohan.r@kpit.com>
rajmohanrgithub
force-pushed
the
patches-2-upstream-systemd
branch
from
9e5a453
to
855b22d
Compare
|
Please set the title of the pull request. |
| @@ -1699,6 +1699,17 @@ static void cmdline_take_random_seed(void) { | |||
| "This functionality should not be used outside of testing environments."); | |||
| } | |||
|
|
|||
| static void set_suid_dumpable(void) { | |||
| const char *sd_path = "/proc/sys/fs/suid_dumpable"; | |||
There was a problem hiding this comment.
Uh, we already do this?
systemd/sysctl.d/50-coredump.conf.in
Line 37 in 1b47cfa
| @@ -2032,6 +2032,14 @@ static int invoke_main_loop( | |||
| assert(ret_switch_root_init); | |||
| assert(ret_error_message); | |||
|
|
|||
| log_warning("sd: i=userspace, inactive_exit="USEC_FMT, m->timestamps[MANAGER_TIMESTAMP_USERSPACE].monotonic); | |||
There was a problem hiding this comment.
If anything, these definitely shouldn't be logged at the warning level (debug seems to be appropriate).
There was a problem hiding this comment.
and they can all be queried at runtime, and some are already logged later, so doesn't seem worth adding any of these
| @@ -1168,6 +1169,16 @@ static void mount_enter_mounting(Mount *m) { | |||
| m->control_command_id = MOUNT_EXEC_MOUNT; | |||
| m->control_command = m->exec_command + MOUNT_EXEC_MOUNT; | |||
|
|
|||
| dp = opendir(m->where); | |||
There was a problem hiding this comment.
I don't think we want to nuke any files from the filesystem which are not under our direct control.
|
Also, please follow https://github.com/systemd/systemd/blob/main/docs/CODING_STYLE.md. |
| @@ -274,6 +276,8 @@ static int help(int argc, char *argv[], void *userdata) { | |||
| " security review of the unit(s)\n" | |||
| " --unit=UNIT Evaluate conditions and asserts of unit\n" | |||
| " --table Output plot's raw time data as a table\n" | |||
| " -s --scale=FACTOR Stretch the x-axis of the plot by FACTOR (default: 1.0)\n" | |||
There was a problem hiding this comment.
shorthands are prime real estate, and sd-analyze has tons of options and will get more. This is not something common enough to warrant taking over one of the 52 available single-letter options, so please just use the long one
bluca
added
the
reviewed/needs-rework 🔨
|
Please split the sd-analyze commit into a separate PR, as that looks reasonable and could be merged quickly once fixed up |
|
Sorry, but as mentioned, all commits except for the sd-analyze one make no sense at all. The timestamps can be queried through dbus, and we do not do the intense logging about them. The mount one is even more problematic. What's the rationale??? |
YHNdnzj
added
pid1
needs-reporter-feedback ❓
Message from commits is given below:
systemd: Add more logs for boot procedure analysis
systemd-analyze: Add svg scaling options
systemd: set suid_dumpable to "suidssafe"
mount: Remove files in place of mount points