oxmc/libgtop
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Yet another kstat_chain_update check. Added machine.cpu_stat_kstat[x] =

Browse Source 
* open.c (glibtop_get_kstats): Yet another kstat_chain_update
        check. Added machine.cpu_stat_kstat[x] = NULL when processor
        x is not configured.

        * procdata.c (glibtop_get_proc_credentials_s): Read prcred
        structure from /proc.

        * procstate.c (glibtop_get_proc_state_s): Added ruid, rgid,
        has_cpu, processor and last_processor.

        * procuid.c (glibtop_get_proc_uid_s): Added priority, nice,
        suid, sgid, ngroups and groups. The last four will be
        filled only if our process has the authority to read prcred
        structure of another process.

	It's a bit untested for now. :-)
This commit is contained in:
Drazen Kacar
1999-05-02 19:26:24 +00:00
parent 3951c8863f
commit 75141bdb65
5 changed files with 168 additions and 64 deletions
Download Patch File Download Diff File Expand all files Collapse all files

115
sysdeps/solaris/open.c
View File

@@ -47,65 +47,76 @@ glibtop_get_kstats(glibtop *server)
server->ncpu = new_ncpu;
server->machine.vminfo_kstat = NULL;
server->machine.system = NULL;
server->machine.syspages = NULL;
server->machine.bunyip = NULL;
return;
}
ksp = kstat_lookup(kc, "unix", -1, "vminfo");
server->machine.vminfo_kstat = ksp;
if(ksp)
{
kstat_read(kc, ksp, &server->machine.vminfo);
server->machine.vminfo_snaptime = ksp->ks_snaptime;
}
/* We don't know why was kstat chain invalidated. It could have
been because the number of processors changed. The sysconf()
man page says that values returned won't change during the
life time of a process, but let's hope that's just an error in
the documentation. */
if(nproc_same = new_ncpu == server->ncpu)
{
int checked, i;
char cpu[20];
for(i = 0, checked = 0; i < GLIBTOP_NCPU || checked == new_ncpu; ++i)
if(server->machine.cpu_stat_kstat[i])
{
sprintf(cpu, "cpu_stat%d", i);
if(!(server->machine.cpu_stat_kstat[i] =
kstat_lookup(kc, "cpu_stat", -1, cpu)))
{
nproc_same = 0;
break;
}
++checked;
}
}
if(!nproc_same)
{
processorid_t p;
int found;
char cpu[20];
if(new_ncpu > GLIBTOP_NCPU)
new_ncpu = GLIBTOP_NCPU;
server->ncpu = new_ncpu;
for(p = 0, found = 0; p < GLIBTOP_NCPU && found != new_ncpu; ++p)
do {
ksp = kstat_lookup(kc, "unix", -1, "vminfo");
server->machine.vminfo_kstat = ksp;
if(ksp)
{
if(p_online(p, P_STATUS) < 0)
continue;
sprintf(cpu, "cpu_stat%d", (int)p);
server->machine.cpu_stat_kstat[p] =
kstat_lookup(kc, "cpu_stat", -1, cpu);
++found;
kstat_read(kc, ksp, &server->machine.vminfo);
server->machine.vminfo_snaptime = ksp->ks_snaptime;
}
}
server->machine.system = kstat_lookup(kc, "unix", -1, "system_misc");
server->machine.syspages = kstat_lookup(kc, "unix", -1, "system_pages");
server->machine.bunyip = kstat_lookup(kc, "bunyip", -1, "mempages");
/* We don't know why was kstat chain invalidated. It could have
been because the number of processors changed. The sysconf()
man page says that values returned won't change during the
life time of a process, but let's hope that's just an error in
the documentation. */
if(nproc_same = new_ncpu == server->ncpu)
{
int checked, i;
char cpu[20];
for(i = 0, checked = 0; i < GLIBTOP_NCPU || checked == new_ncpu; ++i)
if(server->machine.cpu_stat_kstat[i])
{
sprintf(cpu, "cpu_stat%d", i);
if(!(server->machine.cpu_stat_kstat[i] =
kstat_lookup(kc, "cpu_stat", -1, cpu)))
{
nproc_same = 0;
break;
}
++checked;
}
}
if(!nproc_same)
{
processorid_t p;
int found;
char cpu[20];
if(new_ncpu > GLIBTOP_NCPU)
new_ncpu = GLIBTOP_NCPU;
server->ncpu = new_ncpu;
for(p = 0, found = 0; p < GLIBTOP_NCPU && found != new_ncpu; ++p)
{
if(p_online(p, P_STATUS) < 0)
{
server->machine.cpu_stat_kstat[p] = NULL;
continue;
}
sprintf(cpu, "cpu_stat%d", (int)p);
server->machine.cpu_stat_kstat[p] =
kstat_lookup(kc, "cpu_stat", -1, cpu);
++found;
}
}
server->machine.system = kstat_lookup(kc, "unix", -1, "system_misc");
server->machine.syspages = kstat_lookup(kc, "unix", -1, "system_pages");
server->machine.bunyip = kstat_lookup(kc, "bunyip", -1, "mempages");
} while(kstat_chain_update(kc) > 0 &&
(new_ncpu = sysconf(_SC_NPROCESSORS_CONF)));
/* We'll ignore -1 from kstat_chain_update here, since it really
shouldn't happen */
}
void