Bug 26057
| Summary: | [PATCH] Minor ps(1) fixes | ||||||
|---|---|---|---|---|---|---|---|
| Product: | Base System | Reporter: | dd <dd> | ||||
| Component: | bin | Assignee: | freebsd-bugs (Nobody) <bugs> | ||||
| Status: | Closed FIXED | ||||||
| Severity: | Affects Only Me | ||||||
| Priority: | Normal | ||||||
| Version: | Unspecified | ||||||
| Hardware: | Any | ||||||
| OS: | Any | ||||||
| Attachments: |
|
||||||
On Sat, 24 Mar 2001 dd@FreeBSD.ORG wrote: > >Description: > > When the kinfo_proc structure was introduced, the ps(1) manual page > was not updated. Thus, it still talks about keywords that don't > exist. Also, one line in ps.c was forgotten. The latter results in > an annoying warning when using ps(1) with the -j flag (see > how-to-repeat). Some of the bugs are in kinfo_proc, not in ps. > Index: ps.1 > =================================================================== > RCS file: /st/src/FreeBSD/src/bin/ps/ps.1,v > retrieving revision 1.30 > diff -u -r1.30 ps.1 > --- ps.1 2001/02/01 16:24:50 1.30 > +++ ps.1 2001/03/20 03:34:55 > @@ -99,7 +99,7 @@ > header per page of information. > .It Fl j > Print information associated with the following keywords: > -user, pid, ppid, pgid, sess, jobc, state, tt, time and command. > +user, pid, ppid, pgid, jobc, state, tt, time and command. I think the seesion pointer can be used to distinguish sessions. pstat(8) still prints it (pstat gets it in a different way). >... > -.It ktracep > -tracing vnode >... > -.It p_ru > -resource usage (valid only for zombie) Other pointers may be useful for groping in kmem or in panic dumps, but gdb is better for that. I hacked my ps to show the cputime() resource usage for zombies in all cases. It's not very useful to print it as 0. > -.It rsz > -resident set size + (text size / text use count) (alias rssize) Hopefully no one uses this any more. Bruce Bruce Evans <bde@zeta.org.au> writes: > On Sat, 24 Mar 2001 dd@FreeBSD.ORG wrote: > > Index: ps.1 > > =================================================================== > > RCS file: /st/src/FreeBSD/src/bin/ps/ps.1,v > > retrieving revision 1.30 > > diff -u -r1.30 ps.1 > > --- ps.1 2001/02/01 16:24:50 1.30 > > +++ ps.1 2001/03/20 03:34:55 > > @@ -99,7 +99,7 @@ > > header per page of information. > > .It Fl j > > Print information associated with the following keywords: > > -user, pid, ppid, pgid, sess, jobc, state, tt, time and command. > > +user, pid, ppid, pgid, jobc, state, tt, time and command. > > I think the seesion pointer can be used to distinguish sessions. > pstat(8) still prints it (pstat gets it in a different way). This is trivial to fix. See the patch below. > > >... > > -.It ktracep > > -tracing vnode > >... > > -.It p_ru > > -resource usage (valid only for zombie) > > Other pointers may be useful for groping in kmem or in panic dumps, but > gdb is better for that. > > I hacked my ps to show the cputime() resource usage for zombies in all > cases. It's not very useful to print it as 0. struct rusage is exported in kinfo_proc as ki_rusage, but I don't think ps knows about it. I guess this is also trivial to fix, but I haven't tried. I'll supply a patch for this if you want, too. The patch below adds the 'sess' option back to ps. Mckusick explicitly said he was removing it in his kinfo_proc commit, but didn't specify a reason. Regards Dima Dorfman dima@unixfreak.org Index: bin/ps/keyword.c =================================================================== RCS file: /st/src/FreeBSD/src/bin/ps/keyword.c,v retrieving revision 1.29 diff -u -r1.29 keyword.c --- bin/ps/keyword.c 2001/02/14 18:54:34 1.29 +++ bin/ps/keyword.c 2001/03/25 06:35:03 @@ -148,6 +148,7 @@ {"ruid", "RUID", NULL, 0, kvar, NULL, UIDLEN, KOFF(ki_ruid), UINT, UIDFMT}, {"ruser", "RUSER", NULL, LJUST|DSIZ, runame, s_runame, USERLEN}, + {"sess", "SESS", NULL, 0, kvar, NULL, 6, KOFF(ki_sess), KPTR, "lx"}, {"sig", "PENDING", NULL, 0, kvar, NULL, 8, KOFF(ki_siglist), INT, "x"}, {"sigcatch", "CAUGHT", NULL, 0, kvar, NULL, 8, KOFF(ki_sigcatch), UINT, "x"}, Index: lib/libkvm/kvm_proc.c =================================================================== RCS file: /st/src/FreeBSD/src/lib/libkvm/kvm_proc.c,v retrieving revision 1.32 diff -u -r1.32 kvm_proc.c --- lib/libkvm/kvm_proc.c 2001/02/12 00:21:09 1.32 +++ lib/libkvm/kvm_proc.c 2001/03/25 06:35:03 @@ -320,6 +320,7 @@ kp->ki_rqindex = proc.p_rqindex; kp->ki_oncpu = proc.p_oncpu; kp->ki_lastcpu = proc.p_lastcpu; + kp->ki_sess = proc.p_session; bcopy(&kinfo_proc, bp, sizeof(kinfo_proc)); ++bp; ++cnt; Index: sys/kern/kern_proc.c =================================================================== RCS file: /st/src/FreeBSD/src/sys/kern/kern_proc.c,v retrieving revision 1.89 diff -u -r1.89 kern_proc.c --- sys/kern/kern_proc.c 2001/03/07 06:52:12 1.89 +++ sys/kern/kern_proc.c 2001/03/25 06:35:03 @@ -509,6 +509,7 @@ kp->ki_lock = p->p_lock; if (p->p_pptr) kp->ki_ppid = p->p_pptr->p_pid; + kp->ki_sess = p->p_session; PROC_UNLOCK(p); } Index: sys/sys/user.h =================================================================== RCS file: /st/src/FreeBSD/src/sys/sys/user.h,v retrieving revision 1.35 diff -u -r1.35 user.h --- sys/sys/user.h 2001/03/09 11:34:29 1.35 +++ sys/sys/user.h 2001/03/25 06:35:03 @@ -149,7 +149,8 @@ struct rusage ki_rusage; /* process rusage statistics */ long ki_sflag; /* PS_* flags */ struct priority ki_pri; /* process priority */ - long ki_spare[25]; /* spare constants */ + struct session *ki_sess; /* session pointer */ + long ki_spare[24]; /* spare constants */ }; void fill_kinfo_proc __P((struct proc *, struct kinfo_proc *)); State Changed From-To: open->closed I think the reason for the 'w' option in the first place is to disable the default window-width setting. As Wollman notes, breaking a very common case would hurt POLA. <phk@FreeBSD.org> writes: > Synopsis: [PATCH] Minor ps(1) fixes > > State-Changed-From-To: open->closed > State-Changed-By: phk > State-Changed-When: Wed Mar 28 10:27:20 PST 2001 > State-Changed-Why: > I think the reason for the 'w' option in the first place is > to disable the default window-width setting. > > As Wollman notes, breaking a very common case would hurt POLA. Um, I think you closed the wrong PR. This one does not talk about the 'w' option. PR 25855 does. This one fixes this: dd@ref5% ps j ps: sess: keyword not found ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ [ normal ps output ] and some stuff in a man page. Also note that Wollman never said anything about this PR. Please re-open (and/or commit :-) it. Regards Dima Dorfman dima@unixfreak.org State Changed From-To: closed->open Dang, wrong PR. Ignore last entry. Thanks for noticing my cut&paste-o :-! In message <20010328190346.F3D1C3E09@bazooka.unixfreak.org>, Dima Dorfman write s: ><phk@FreeBSD.org> writes: >> Synopsis: [PATCH] Minor ps(1) fixes >> >> State-Changed-From-To: open->closed >> State-Changed-By: phk >> State-Changed-When: Wed Mar 28 10:27:20 PST 2001 >> State-Changed-Why: >> I think the reason for the 'w' option in the first place is >> to disable the default window-width setting. >> >> As Wollman notes, breaking a very common case would hurt POLA. > >Um, I think you closed the wrong PR. This one does not talk about the >'w' option. PR 25855 does. This one fixes this: > > dd@ref5% ps j > ps: sess: keyword not found > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > [ normal ps output ] > >and some stuff in a man page. Also note that Wollman never said >anything about this PR. > >Please re-open (and/or commit :-) it. > >Regards > > Dima Dorfman > dima@unixfreak.org > > -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 phk@FreeBSD.ORG | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. State Changed From-To: open->closed ps.c part committed by brian (1.37 of bin/ps/ps.c). I'll update the man page later today. On Fri, 6 Apr 2001 dd@FreeBSD.ORG wrote: > Synopsis: [PATCH] Minor ps(1) fixes > > State-Changed-From-To: open->closed > State-Changed-By: dd > State-Changed-When: Fri Apr 6 20:19:09 PDT 2001 > State-Changed-Why: > ps.c part committed by brian (1.37 of bin/ps/ps.c). I'll update the > man page later today. The committed fix is half-baked at best. It does't even remove the session keyword like the first version of the PR does. Kirk also removed the tsess pointer and keyword from everywhere except ps.1, but the PR doesn't mention it. The log message for rev.1.25 of keyword.c says the following about the removal: kernel/user interface, specifically rusage and rtprio. It no longer contains proc, session, pcred, ucred, procsig, vmspace, pstats, mtx, sigiolst, klist, callout, pasleep, or mdproc. If any of these changed in size, ps, w, fstat, gcore, systat, and top would all stop working. The new structure has over 200 bytes This is incorrect for "session" in ps at least. ps would not have stopped working if sizeof(struct session) changed, since it just printed the session pointer. Bruce |
When the kinfo_proc structure was introduced, the ps(1) manual page was not updated. Thus, it still talks about keywords that don't exist. Also, one line in ps.c was forgotten. The latter results in an annoying warning when using ps(1) with the -j flag (see how-to-repeat). Fix: The following patch fixes the warning and updates the man page. I'm only a doc committer, so someone else needs to commit the src/ part. How-To-Repeat: dd@ref5% ps -j ps: sess: keyword not found [ normal ps(1) output follows ]