RE: Using serial console to debug system hangs ...

2001-03-04 Thread John Baldwin


On 04-Mar-01 The Hermit Hacker wrote:
 On Sat, 3 Mar 2001, John Baldwin wrote:
 

 On 04-Mar-01 The Hermit Hacker wrote:
 
  Wow, that was painful ... after 2 hrs, I got as far as:

 Yeah, it spews out a lot of crap. :-/  You prolly want to use a 115200
 serial
 console if at all possible.  Should've mentioned that earlier..
 
 Okay, reading NOTES, it says that 9600 is the default ... does that mean I
 should be able to attach at 115200 and it should auto-upgrade, or do I
 have to recompile kernel iwth CONSPEED=115200 for this?

Either recompile or use the loader tunable 'machdep.conspeed' I think.
However, you'll probably want the bootstrap to work on teh serial console as
well, in which case you need to set the speed in make.conf (see
/etc/defaults/make.conf) and recompile and reinstall boot2 and the loader. 

-- 

John Baldwin [EMAIL PROTECTED] -- http://www.FreeBSD.org/~jhb/
PGP Key: http://www.baldwin.cx/~john/pgpkey.asc
"Power Users Use the Power to Serve!"  -  http://www.FreeBSD.org/

To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message



RE: Using serial console to debug system hangs ...

2001-03-04 Thread The Hermit Hacker

On Sun, 4 Mar 2001, John Baldwin wrote:


 On 04-Mar-01 The Hermit Hacker wrote:
  On Sat, 3 Mar 2001, John Baldwin wrote:
 
 
  On 04-Mar-01 The Hermit Hacker wrote:
  
   Wow, that was painful ... after 2 hrs, I got as far as:
 
  Yeah, it spews out a lot of crap. :-/  You prolly want to use a 115200
  serial
  console if at all possible.  Should've mentioned that earlier..
 
  Okay, reading NOTES, it says that 9600 is the default ... does that mean I
  should be able to attach at 115200 and it should auto-upgrade, or do I
  have to recompile kernel iwth CONSPEED=115200 for this?

 Either recompile or use the loader tunable 'machdep.conspeed' I think.
 However, you'll probably want the bootstrap to work on teh serial console as
 well, in which case you need to set the speed in make.conf (see
 /etc/defaults/make.conf) and recompile and reinstall boot2 and the loader.

Okay, now you are going beyond my experiences :(

First, I changed CONSPEED to 115200, but sysctl is still registering 9600:

thelab# sysctl machdep.conspeed
machdep.conspeed: 9600

manually doing 'sysctl -w machdep.conspeed=115200' appears to work though
...

But, after modifying my make.conf, what do I have to recompile?  Are we
just talking a rebuild and install of the kernel itself?  I'm going to try
that ...



To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message



RE: Using serial console to debug system hangs ...

2001-03-04 Thread Manfred Antar

At 01:48 PM 3/4/2001 -0400, The Hermit Hacker wrote:
On Sun, 4 Mar 2001, John Baldwin wrote:


 On 04-Mar-01 The Hermit Hacker wrote:
  On Sat, 3 Mar 2001, John Baldwin wrote:
 
 
  On 04-Mar-01 The Hermit Hacker wrote:
  
   Wow, that was painful ... after 2 hrs, I got as far as:
 
  Yeah, it spews out a lot of crap. :-/  You prolly want to use a 115200
  serial
  console if at all possible.  Should've mentioned that earlier..
 
  Okay, reading NOTES, it says that 9600 is the default ... does that mean I
  should be able to attach at 115200 and it should auto-upgrade, or do I
  have to recompile kernel iwth CONSPEED=115200 for this?

 Either recompile or use the loader tunable 'machdep.conspeed' I think.
 However, you'll probably want the bootstrap to work on teh serial console as
 well, in which case you need to set the speed in make.conf (see
 /etc/defaults/make.conf) and recompile and reinstall boot2 and the loader.

Okay, now you are going beyond my experiences :(

First, I changed CONSPEED to 115200, but sysctl is still registering 9600:

thelab# sysctl machdep.conspeed
machdep.conspeed: 9600

manually doing 'sysctl -w machdep.conspeed=115200' appears to work though
...

But, after modifying my make.conf, what do I have to recompile?  Are we
just talking a rebuild and install of the kernel itself?  I'm going to try
that ...



To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message 

You have to recompile the boot stuff also, after changing the line in make.conf:
# The default serial console speed is 9600.  Set the speed to a larger value
# for better interactive response.
#
BOOT_COMCONSOLE_SPEED=  57600

Then cd /sys/boot ; make depend all install.
I forget if you then need to relabel the disk or not.
ie :
disklabel -B da0

==
||  [EMAIL PROTECTED]   ||
||  Ph. (415) 681-6235  ||
==


To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message



RE: Using serial console to debug system hangs ...

2001-03-04 Thread The Hermit Hacker

On Sun, 4 Mar 2001, Manfred Antar wrote:

 You have to recompile the boot stuff also, after changing the line in make.conf:
 # The default serial console speed is 9600.  Set the speed to a larger value
 # for better interactive response.
 #
 BOOT_COMCONSOLE_SPEED=  57600

 Then cd /sys/boot ; make depend all install.
 I forget if you then need to relabel the disk or not.
 ie :
 disklabel -B da0

this worked great, thanks ...disklabel wasn't required, it appears ...



To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message



RE: Using serial console to debug system hangs ...

2001-03-04 Thread The Hermit Hacker

On Sat, 3 Mar 2001, John Baldwin wrote:


 On 04-Mar-01 The Hermit Hacker wrote:
 
  Wow, that was painful ... after 2 hrs, I got as far as:

 Yeah, it spews out a lot of crap. :-/  You prolly want to use a 115200 serial
 console if at all possible.  Should've mentioned that earlier..

Okay, I'm up to 115200, still a lot of crap ;)  Any lower mask I can set
things at to get enough info, without so much flowing up the screen?  I
can understand why she's so unresponsive, just wondering if there is a way
of reducing the amount of debugging without losing too much ...



To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message



RE: Using serial console to debug system hangs ...

2001-03-04 Thread John Baldwin


On 04-Mar-01 The Hermit Hacker wrote:
 On Sun, 4 Mar 2001, Manfred Antar wrote:
 
 You have to recompile the boot stuff also, after changing the line in
 make.conf:
 # The default serial console speed is 9600.  Set the speed to a larger value
 # for better interactive response.
 #
 BOOT_COMCONSOLE_SPEED=  57600

 Then cd /sys/boot ; make depend all install.
 I forget if you then need to relabel the disk or not.
 ie :
 disklabel -B da0
 
 this worked great, thanks ...disklabel wasn't required, it appears ...

It is to update boot2.  However, you don't have to do that if you don't want.
You can just boot on a normal console then break into the loader and type
'set console=comconsole' to switch over to the serial console.

-- 

John Baldwin [EMAIL PROTECTED] -- http://www.FreeBSD.org/~jhb/
PGP Key: http://www.baldwin.cx/~john/pgpkey.asc
"Power Users Use the Power to Serve!"  -  http://www.FreeBSD.org/

To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message



RE: Using serial console to debug system hangs ...

2001-03-04 Thread John Baldwin


On 04-Mar-01 The Hermit Hacker wrote:
 On Sat, 3 Mar 2001, John Baldwin wrote:
 

 On 04-Mar-01 The Hermit Hacker wrote:
 
  Wow, that was painful ... after 2 hrs, I got as far as:

 Yeah, it spews out a lot of crap. :-/  You prolly want to use a 115200
 serial
 console if at all possible.  Should've mentioned that earlier..
 
 Okay, I'm up to 115200, still a lot of crap ;)  Any lower mask I can set
 things at to get enough info, without so much flowing up the screen?  I
 can understand why she's so unresponsive, just wondering if there is a way
 of reducing the amount of debugging without losing too much ...

Well.  Currnetly what 0x1200 logs is KTR_PROC and KTR_INTR, so it logs all
incoming threaded interrupts as well as context switches, etc.  If the machine
hagns, it is very helpful to know if a) we are still getting interrupts at all
(such as clock interrupts, which I can tell beecause the softclock swi thread
get scheduled to execute.) and b) what processes were doing what when the
machien hung.

-- 

John Baldwin [EMAIL PROTECTED] -- http://www.FreeBSD.org/~jhb/
PGP Key: http://www.baldwin.cx/~john/pgpkey.asc
"Power Users Use the Power to Serve!"  -  http://www.FreeBSD.org/

To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message



Re: Using serial console to debug system hangs ...

2001-03-04 Thread Warner Losh

In message [EMAIL PROTECTED] Manfred Antar writes:
: Then cd /sys/boot ; make depend all install.
: I forget if you then need to relabel the disk or not.
: ie :
: disklabel -B da0

I've noticed in the past that FreeBSD uss the com speed of the boot
blocks.

Warner


To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message



Re: Using serial console to debug system hangs ...

2001-03-04 Thread Manfred Antar

At 05:49 PM 3/4/2001 -0700, Warner Losh wrote:
In message [EMAIL PROTECTED] Manfred Antar writes:
: Then cd /sys/boot ; make depend all install.
: I forget if you then need to relabel the disk or not.
: ie :
: disklabel -B da0

I've noticed in the past that FreeBSD uss the com speed of the boot
blocks.

Warner

Yes
I thought so. There is a web page some where that tell's exactly how to set this up.
I just found it:

http://www.freebsd.org:80/handbook/serialconsole-setup.html

Manfred
==
||  [EMAIL PROTECTED]   ||
||  Ph. (415) 681-6235  ||
==


To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message



RE: Using serial console to debug system hangs ...

2001-03-03 Thread John Baldwin


On 04-Mar-01 The Hermit Hacker wrote:
 
 Well, after some hurdles with getting the serial console to work, I've now
 go it to work ... I put the two sysctl commands into a file so that I
 could run it as a script:
 
#!/bin/sh
 sysctl -w debug.ktr.mask=0x1208
 sysctl -w debug.ktr.verbose=2
 
 When I 'try' to run it, I get all the 'KTR'(?) messages on my serial
 console, something is changing/happening so fast that my ssh connection
 into the machine hangs before I finish typing in the shell script:
 
 
 enable_kernel_debug: 3 lines, 72 characters.
 thelab# !./
 

Put the sysctl's and the command that hangs the machine into one script and run
that one shell script..

 needless to say, running the command to hang the computer is proving
 difficult :)
 
 Then again, if I do a cold boot of the machine, the messages stop
 scrolling up the console, but a cut-n-paste of them is sort of illegible:
 
 
  k0clo.c/k...c/.k4e3r8n /RkEeLr n(_scplionc)k .scch2e0d9 lRoEcLk 
 (s[p0xicn0) 32c1al1l8o0]ut  r[=00x ca0t 31.d./8.2.0/] ker=r0n/ akte
 r.n._c/l.o./ckk.ecrn:4/k3e8r
 e_cpluo1c k..c/:.2.0/9k
  rcnp/uk0e r.n./.l.o/ckke.rcn./3k5e0r   nG_OcTl o(cskp.icn.)2 0s3c 
 hGeOdT  l(oscpki n[)0 xccall2o1u1t8 0[]0 xrc=003 1adt8 2.0.]/ ..0/k aertn
 ./.k/er..n_/ckllorcnk/.kecr:n35_0c
 .ok.uc1: 2.0.3/
  .c/pkue0r.n/.k/.er./n_kcelrnoc/kke.crn.4_c38l  ocRkEL. c(.s2p09in  )R
 ELs ch(sepd iln)oc kc al[0loxuct03 [210x18c003] 1rd=802 0a] t r=..0/ a..t
 /.k.er/.n./.ckerenrn_c/lkoerckn_.ccl:4oc3k8
   :20u91
  .c.p/u.0. /.k.e/r.n.//kkeerrnn_/ckleorcnk_.ccl.o3c5k0. cG.O2T0 3( 
 sGpOiTn )( sspcihne)d  claolclko u[t0 x[c00x3213118d08]2 0r]= 0r =a0t  a.t.
 ///.k.e/0ke/rkne/rkne_rcnl_occlko.cck:.3c5:02
 p3c
  uc1p u.0. ///.k.e/rkne/rkne/rkne_rcnl_occlko.cck..4c3.82   RE  LR
 E(Ls p(isnp)i ns)c hceadl lloouctk  [[00xxcc003312d1812800]]  rr==00  aatt 
 ////kkeerrnn//kkeerrnn__cclloocckk..cc::24398
 
 ccppuu01  ////kkeerrnn//kkeerrnn__cclloocckk..cc..230530 
 GGOOTT  ((ssppiinn))  csaclhleodu tl o[c0kx c[003x1cd0832201]1 8r0=]0  ra=t0 
 .a.t
 

Hmm, it's colliding with itself a lot.  Unfortunately, to make this useful over
the serial console, you need to shut up all the sio lock messages.  Hmmm, well
for now try just using a 'debug.ktr.mask' of 0x1200 to skip all the mutex
operations.  If we need them later on, then I will try and get some other work
done to make it easier to shut up certain mutexes in the log output without
having to change each individual mutex operation.

-- 

John Baldwin [EMAIL PROTECTED] -- http://www.FreeBSD.org/~jhb/
PGP Key: http://www.baldwin.cx/~john/pgpkey.asc
"Power Users Use the Power to Serve!"  -  http://www.FreeBSD.org/

To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message



RE: Using serial console to debug system hangs ...

2001-03-03 Thread The Hermit Hacker


Wow, that was painful ... after 2 hrs, I got as far as:

thelab# ./enable_debug
debug.ktr.mask: 1 - 4608
debug.ktr.verbose: 0 - 2


waiting for X server to begin accepting connections .

XFree86 Version 4.0.2 / X Window System
(protocol Version 11, revision 0, vendor release 6400)
Release Date: 18 December 2000
If the server is older than 6-12 months, or if your card is
newer than the above date, look for a newer version before
reporting problems.  (See http://www.XFree86.Org/FAQ)
Operating System: FreeBSD 5.0-CURRENT i386 [ELF]
Module Loader present
(==) Log file: "/var/log/XFree86.0.log", Time: Sat Mar  3 22:08:37 2001
(==) Using config file: "/root/XF86Config"
Markers: (--) probed, (**) from config file, (==) default setting,
 (++) from command line, (!!) notice, (II) informational,
 (WW) warning, (EE) error, (??) unknown.
(==) ServerLayout "Simple Layout"
(**) |--Screen "Screen 1" (0)
(**) |   |--Monitor "Samtron 95P"
(**) |   |--Device "TNT"
(**) |--Input Device "Mouse1"
(**) |--Input Device "Keyboard1"
(**) XKB: rules: "xfree86"
(**) XKB: model: "microsoft"
(**) XKB: layout: "us"
(**) FontPath set to 
"/usr/X11R6/lib/X11/fonts/local/,/usr/X11R6/lib/X11/fonts/misc/,/usr/X11R6/lib/X11/fonts/75dpi/:unscaled,/usr/X11R6/lib/X11/fonts/100dpi/:unscaled,/usr/X11R6/lib/X11/fonts/Type1/,/usr/X11R6/lib/X11/fonts/Speedo/,/usr/X11R6/lib/X11/fonts/75dpi/,/usr/X11R6/lib/X11/fonts/100dpi/"
(**) RgbPath set to "/usr/X11R6/lib/X11/rgb"
(==) ModulePath set to "/usr/X11R6/lib/modules"
(--) Using syscons driver with X support (version 2.0)

and gave up ... am going to try again just before I go to bed tonight,
hopefully its running (and hung) by the time I get up in the morning ;)




On Sat, 3 Mar 2001, John Baldwin wrote:


 On 04-Mar-01 The Hermit Hacker wrote:
 
  Well, after some hurdles with getting the serial console to work, I've now
  go it to work ... I put the two sysctl commands into a file so that I
  could run it as a script:
 
 #!/bin/sh
  sysctl -w debug.ktr.mask=0x1208
  sysctl -w debug.ktr.verbose=2
 
  When I 'try' to run it, I get all the 'KTR'(?) messages on my serial
  console, something is changing/happening so fast that my ssh connection
  into the machine hangs before I finish typing in the shell script:
 
  
  enable_kernel_debug: 3 lines, 72 characters.
  thelab# !./
  

 Put the sysctl's and the command that hangs the machine into one script and run
 that one shell script..

  needless to say, running the command to hang the computer is proving
  difficult :)
 
  Then again, if I do a cold boot of the machine, the messages stop
  scrolling up the console, but a cut-n-paste of them is sort of illegible:
 
  
   k0clo.c/k...c/.k4e3r8n /RkEeLr n(_scplionc)k .scch2e0d9 lRoEcLk
  (s[p0xicn0) 32c1al1l8o0]ut  r[=00x ca0t 31.d./8.2.0/] ker=r0n/ akte
  r.n._c/l.o./ckk.ecrn:4/k3e8r
  e_cpluo1c k..c/:.2.0/9k
   rcnp/uk0e r.n./.l.o/ckke.rcn./3k5e0r   nG_OcTl o(cskp.icn.)2 0s3c
  hGeOdT  l(oscpki n[)0 xccall2o1u1t8 0[]0 xrc=003 1adt8 2.0.]/ ..0/k aertn
  ./.k/er..n_/ckllorcnk/.kecr:n35_0c
  .ok.uc1: 2.0.3/
   .c/pkue0r.n/.k/.er./n_kcelrnoc/kke.crn.4_c38l  ocRkEL. c(.s2p09in  )R
  ELs ch(sepd iln)oc kc al[0loxuct03 [210x18c003] 1rd=802 0a] t r=..0/ a..t
  /.k.er/.n./.ckerenrn_c/lkoerckn_.ccl:4oc3k8
:20u91
   .c.p/u.0. /.k.e/r.n.//kkeerrnn_/ckleorcnk_.ccl.o3c5k0. cG.O2T0 3(
  sGpOiTn )( sspcihne)d  claolclko u[t0 x[c00x3213118d08]2 0r]= 0r =a0t  a.t.
  ///.k.e/0ke/rkne/rkne_rcnl_occlko.cck:.3c5:02
  p3c
   uc1p u.0. ///.k.e/rkne/rkne/rkne_rcnl_occlko.cck..4c3.82   RE  LR
  E(Ls p(isnp)i ns)c hceadl lloouctk  [[00xxcc003312d1812800]]  rr==00  aatt
  ////kkeerrnn//kkeerrnn__cclloocckk..cc::24398
 
  ccppuu01  ////kkeerrnn//kkeerrnn__cclloocckk..cc..230530
  GGOOTT  ((ssppiinn))  csaclhleodu tl o[c0kx c[003x1cd0832201]1 8r0=]0  ra=t0
  .a.t
  

 Hmm, it's colliding with itself a lot.  Unfortunately, to make this useful over
 the serial console, you need to shut up all the sio lock messages.  Hmmm, well
 for now try just using a 'debug.ktr.mask' of 0x1200 to skip all the mutex
 operations.  If we need them later on, then I will try and get some other work
 done to make it easier to shut up certain mutexes in the log output without
 having to change each individual mutex operation.

 --

 John Baldwin [EMAIL PROTECTED] -- http://www.FreeBSD.org/~jhb/
 PGP Key: http://www.baldwin.cx/~john/pgpkey.asc
 "Power Users Use the Power to Serve!"  -  http://www.FreeBSD.org/


Marc G. Fournier   ICQ#7615664   IRC Nick: Scrappy
Systems Administrator @ hub.org
primary: [EMAIL PROTECTED]   secondary: scrappy@{freebsd|postgresql}.org


To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message