Recent changes to patches

#7 Fix incorrect error message array

2021-02-08T10:01:15.715000Z

This may lead to returning invalid error message from cgroup_strerror()
or accessing out of archive boundary when cgroup_strerror callew with
code=ECGNONEMPTY.

Fix incorrect error message array

2021-02-08T09:43:32.115000Z

cgroup_strerror_codes in src/api.c does not match to ECG* error enum defined in include/libcgroup/error.h
I have added one more error message to fill the gap between ECGMOUNTFAIL (50021 - Cgroup mounting failed) and ECGEOF (50023 - End of File or iterator) messages.

#5 Increasing the value of CG_VALUE_MAX

2019-07-26T19:12:54.989000Z

I have also encountered this issue on a machine with 72 cores and alternating cpu ids on each NUMA node (even cpus on numa0, odd on numa1). The value gets truncated after a comma and fails validation. Setting CG_VALUE_MAX to a larger value fixes it.

For reference:
The cpuset:
1,3,5,7,9,11,13,15,17,19,21,23,25,27,29,31,33,35,37,39,41,43,45,47,49,51,53,55,57,59,61,63,65,67,69,71

gets truncated to:
1,3,5,7,9,11,13,15,17,19,21,23,25,27,29,31,33,35,37,39,41,43,45,47,49,51,53,55,57,59,61,63,65,67,69,

This is especially bad because if it were to truncate at the final comma, the error would be invisible to the user.

#6 Misc bugfixes and features

2016-12-06T19:44:25.626000Z

I do not see any bug reports with memleaks or infinite loops (mostly in %u templates), nor feaure requests that this patch covers, so perhaps we could just leave it here for now? If anyone ever needs it or project returns to active state, I can split it up. I haven't touched that code for over a year. :P

#6 Misc bugfixes and features

2016-12-06T18:56:41.626000Z

It has been a while since we have done any active development on libcgroup. If you are willing, I can review patches. Can you split it up, and send it to libcg-devel@lists.sourceforge.net with the kernel DCO being followed?

Thanks!
Dhaval

#6 Misc bugfixes and features

2016-12-06T18:53:42.717000Z

Without SF fromatting:

-b <size> | --netlink-buffer=<size> set the netlink buffer size in bytes
-p <type> | --procname-fallback=<type> set a default fallback for
                   | process name. Type can be one of "exe" (default) or "status"
-k        | --kernel-procname   set a name of kernel processes to "KERNEL"

Misc bugfixes and features

2016-12-06T18:46:37.472000Z

Hi! I just wanted to drop a patch for cgroup daemon that I wrote and used for some time. It contains a few bugfixes (including memleak and infinite loop) and 3 new command line options:

-b <size>	--netlink-buffer=<size> set the netlink buffer size in bytes
-k	--kernel-procname set a name of kernel processes to "KERNEL"

There is also one optimization for busy systems, which might look a bit ugly. Sorry for uploading it raw in one big package, but perhaps someone will find some parts usefull.

Increasing the value of CG_VALUE_MAX

2014-06-26T21:19:12.809000Z

CG_VALUE_MAX is arbitrarily set to 100 which limits string values to a max of 100 characters. A potential situation where this will cause a failure is in the case of setting cpuset.cpus to a long list (unrolled/comma-separated) of logical cpu ids.

Example: Passing a list of 64 logical cpu ids as a string

"0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63"

will truncate and result in a badly formatted string value that causes an EOF

"0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36\x01"

This patch increases CG_VALUE_MAX to a value of 1024. While still somewhat arbitrary, it seems this value should cover even the largest case. And since it looks like the string value buffers are being statically allocated per thread, it shouldn't really result in any major impact in initial memory.

support indented comments

2012-10-25T09:26:30Z

reproduce:
- edit cgconfig.conf
- add a comment line preceeded by spaces [ \t]+
- run cgconfigparser

expected:
- works

instead:
- doesn't recognize the comment line

Fix cgroup_get_cgroup

2012-09-21T13:37:46Z

Currently (0.39-rc1), cgroup_get_cgroup applied on a cgroup with controllers and parameters fails. This patch allow the use of cgroup_get_cgroup with non-empty cgroup, overriding existing values.