From e7bfa6cb1bb043d6ea66e03b32981e26965d2da3 Mon Sep 17 00:00:00 2001
From: Glauber Costa <glommer@cloudius-systems.com>
Date: Thu, 24 Apr 2014 11:32:28 +0400
Subject: [PATCH] linux: provide a /proc/$$/stat compatibility file
MongoDB wants it. In general, I am providing the information that is easy
to get, and ignoring the ones which are not - with the exception of process
count, that seemed easy enough to implement.
This is the kind of thing Mongo does with it:
2014-04-15T09:54:12.322+0000 [clientcursormon] mem (MB) res:670160 virt:25212
2014-04-15T09:54:12.323+0000 [clientcursormon] mapped (incl journal view):160
2014-04-15T09:54:12.324+0000 [clientcursormon] connections:0
Signed-off-by: Glauber Costa <glommer@cloudius-systems.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>
---
core/sched.cc | 6 ++++
fs/procfs/procfs_vnops.cc | 73 +++++++++++++++++++++++++++++++++++++++
include/osv/sched.hh | 2 ++
3 files changed, 81 insertions(+)
diff --git a/core/sched.cc b/core/sched.cc
index 7bf9e0555..a8b7aa034 100644
--- a/core/sched.cc
+++ b/core/sched.cc
@@ -568,6 +568,12 @@ std::chrono::nanoseconds osv_run_stats()
return std::chrono::duration_cast<std::chrono::nanoseconds>(total_app_time);
}
+int thread::numthreads()
+{
+ SCOPE_LOCK(thread_map_mutex);
+ return thread_map.size();
+}
+
// We reserve a space in the end of the PID space, so we can reuse those
// special purpose ids for other things. 4096 positions is arbitrary, but
// <<should be enough for anybody>> (tm)
diff --git a/fs/procfs/procfs_vnops.cc b/fs/procfs/procfs_vnops.cc
index 77839072d..7fed5d09c 100644
--- a/fs/procfs/procfs_vnops.cc
+++ b/fs/procfs/procfs_vnops.cc
@@ -17,6 +17,12 @@
#include <map>
#include "fs/vfs/vfs.h"
+#include <libgen.h>
+#include <osv/mempool.hh>
+#include <osv/printf.hh>
+
+#include <sys/resource.h>
+
namespace procfs {
using namespace std;
@@ -255,6 +261,71 @@ procfs_readdir(vnode *vp, file *fp, dirent *dir)
return 0;
}
+static std::string procfs_stats()
+{
+ int pid = 0, ppid = 0, pgrp = 0, session = 0, tty = 0, tpgid = -1,
+ flags = 0;
+ // Postpone this one. We need to hook into ZFS statistics to properly figure
+ // out which faults are maj, which are min.
+ int min_flt = 0, cmin_flt = 0, maj_flt = 0, cmaj_flt = 0;
+ unsigned long stime = 0, cutime = 0, cstime = 0;
+
+ using namespace std::chrono;
+ unsigned long utime = duration_cast<microseconds>(sched::osv_run_stats()).count();
+ char state = 'R';
+ int priority = getpriority(PRIO_PROCESS, 0);
+ int nice = priority;
+ int nlwp = sched::thread::numthreads();
+ int start_time = 0; // boot is reference, so it is 0.
+ unsigned long rss = memory::stats::total() - memory::stats::free();
+ unsigned long vsize = mmu::all_vmas_size();
+ unsigned long rss_rlim = memory::stats::total();
+ void *start_code = nullptr;
+ void *end_code = nullptr;
+ void *start_stack = nullptr;
+ void *kstk_esp = nullptr;
+ void *kstk_eip = nullptr;
+
+ // All the signal stuff is marked in the manpage as obsolete.
+ unsigned long pending = 0;
+ unsigned long blocked = 0;
+ unsigned long sigign = 0;
+ unsigned long sigcatch = 0;
+ unsigned long nextalarm = 0;
+ // Except for this. This is maintained, but we also don't deliver any signal
+ unsigned long exit_signal = 0;
+ int cpu = sched::cpu::current()->id;
+ int wchan = 0, rt_priority = 0, policy = 0; // SCHED_OTHER = 0
+ int zero = 0;
+
+ std::ostringstream os;
+ osv::fprintf(os, "%d (%s) %c "
+ "%d %d %d %d %d "
+ "%lu %lu %lu %lu %lu "
+ "%lu %lu %ld %ld "
+ "%ld %ld %ld %ld "
+ "%lu %lu "
+ "%lu %lu "
+ "%p %p %p %p %p "
+ "%lu %lx %lu %lu "
+ "%d %d %d "
+ "%lu %d %d %d",
+ pid, program_invocation_short_name, state,
+ ppid, pgrp, session, tty, tpgid,
+ flags, min_flt, cmin_flt, maj_flt, cmaj_flt,
+ utime, stime, cutime, cstime,
+ priority, nice, nlwp, nextalarm,
+ start_time, vsize,
+ rss, rss_rlim,
+ start_code, end_code, start_stack, kstk_esp, kstk_eip,
+ pending, blocked, sigign, sigcatch,
+ wchan, zero, zero,
+ exit_signal, cpu, rt_priority, policy
+ );
+ return os.str();
+
+}
+
static int
procfs_mount(mount* mp, char *dev, int flags, void* data)
{
@@ -262,9 +333,11 @@ procfs_mount(mount* mp, char *dev, int flags, void* data)
auto self = make_shared<proc_dir_node>(inode_count++);
self->add("maps", inode_count++, mmu::procfs_maps);
+ self->add("stat", inode_count++, procfs_stats);
auto* root = new proc_dir_node(vp->v_ino);
root->add("self", self);
+ root->add("0", self); // our standard pid
vp->v_data = static_cast<void*>(root);
diff --git a/include/osv/sched.hh b/include/osv/sched.hh
index becd9639d..befa5d0e1 100644
--- a/include/osv/sched.hh
+++ b/include/osv/sched.hh
@@ -572,6 +572,8 @@ public:
lockless_queue_link<thread> _wakeup_link;
static unsigned long _s_idgen;
static thread *find_by_id(unsigned int id);
+
+ static int numthreads();
private:
class reaper;
friend class reaper;
--
GitLab