On 10/29/2021 7:31 AM, Michal Prívozník wrote:
On 10/22/21 5:37 PM, Praveen K Paladugu wrote:
From: Vineeth Pillai <viremana@xxxxxxxxxxxxxxxxxxx>
These helper methods are used to capture vcpu information
in ch driver.
Signed-off-by: Vineeth Pillai <viremana@xxxxxxxxxxxxxxxxxxx>
Signed-off-by: Praveen K Paladugu <prapal@xxxxxxxxxxxxxxxxxxx>
---
src/util/virprocess.c | 136 ++++++++++++++++++++++++++++++++++++++++++
src/util/virprocess.h | 5 ++
2 files changed, 141 insertions(+)
diff --git a/src/util/virprocess.c b/src/util/virprocess.c
index 6de3f36f52..0164d70df6 100644
--- a/src/util/virprocess.c
+++ b/src/util/virprocess.c
@@ -1721,3 +1721,139 @@ virProcessSetScheduler(pid_t pid G_GNUC_UNUSED,
}
#endif /* !WITH_SCHED_SETSCHEDULER */
+
+/*
+TODO: This method was cloned from qemuGetProcessInfo in src/qemu/qemu_driver.c.
+Need to refactor qemu driver to use this shared function.
There's no harm in doing that in this patch. In fact, it's desired. You
can move a qemu function into src/util, rename it and fix all places
where it is called, all in one patch.
Also, please don't forget to export this function in
src/libvirt_private.syms.
I am working on this now. Will make this part of next version.
+*/
+int
+virProcessGetStatInfo(unsigned long long *cpuTime, int *lastCpu, long *vm_rss,
+ pid_t pid, pid_t tid)
Indentation's off. I've noticed other patches in the series suffer from
mis-indentation too. Please fix that in another version.
Regarding indentation, many of these patches add/modify existing files.
Running indent will modify sections of the files not relevant to the
code changes in this patch set.
Should I run "indent" either way? Or should I make all indentation
changes for all files as a single commit?
+{
+ g_autofree char *proc = NULL;
+ FILE *pidinfo;
+ unsigned long long usertime = 0, systime = 0;
I wonder whether we should take this opportunity and declare these two
variables at one line each.
+ long rss = 0;
+ int cpu = 0;
Michal
--
Regards,
Praveen K Paladugu