Skip to contentSkip to conversation

2.4.23-ck1

DISCUSSION Con Kolivas 8 participants 12 messages Dec 3, 2003 – Dec 4, 2003 Proc filesystem

lore
Share & tools
lore
#0 lore raw

Updated my patchset.

http://kernel.kolivas.org

Includes:
O(1) scheduler with batch scheduling, interactivity Preemptible kernel
Low Latency
Read Latency2
Variable Hz
64bit jiffies
Supermount-NG v1.2.10
Bootsplash v3.0.7
XFS file system v1.3.1

Added:
64 bit jiffies - too many people were complaining of their uptime resetting after 49.7 days at 1000Hz (a good sign I guess) so I've added this patch from Robert Love
ACL and DMAPI for XFS

Updated:
Low latency points that were missing from reiserfs Supermount-NG to latest version
Bootsplash to latest version
XFS to latest stable version.
Variable Hz is set to 1000 by default for all settings.

Removed:
AA VM hacks; they are now in vanilla kernel
GRSecurity: this will have to be a separate patch

Pending (maybe):
GRsec update
rmap option

Notes.
XFS is compile tested only. Test carefully before using in production environment.

I am no longer taking requests for more patches in this patchset, but I will accept sent patches/fixes. This release is mainly a resync and I don't have time for kernel development any more.

Con

#3 lore raw

Hi Marius Ascheberg,

Quoted 9 lines · On Wed, 3 Dec 2003 18:46:34 +0100 Marius Ascheberg <[email protected]> wrote:
Quoted 6 lines · Norberto Bensa
Quoted 3 lines · Con Kolivas

Updated my patchset.

http://kernel.kolivas.org

Too bad http://ck.kolivas.org/ seems down :-(

Con's site is also available through http://members.optusnet.com.au/ckolivas/kernel/ .

Yep but no way to download any patch.. Is this an official mirror from C. Kolivas (that doesn't mirror the files or I'm wrong)?

Regards,

Signature · 1 line
wwp
#4 lore raw
In reply to Marius Ascheberg
Quoted 11 lines · Marius Ascheberg
Quoted 6 lines · Norberto Bensa
Quoted 3 lines · Con Kolivas

Updated my patchset.

http://kernel.kolivas.org

Too bad http://ck.kolivas.org/ seems down :-(

Con's site is also available through http://members.optusnet.com.au/ckolivas/kernel/ .

Marius.

The problem isn't that we can't get to the website, we can't download anything. ck.kolivas.org is where the patches are downloaded from.

Signature · 6 lines
Josh McKinney		     |	Webmaster: http://joshandangie.org
--------------------------------------------------------------------------
                             | They that can give up essential liberty
Linux, the choice       -o)  | to obtain a little temporary safety deserve 
of the GNU generation    /\  | neither liberty or safety. 
                        _\_v |                          -Benjamin Franklin
#6 lore raw
In reply to Con Kolivas
Quoted 1 line · Con Kolivas

Updated my patchset.

There is something wrong with /proc/stat which confuses xosview and ksysguard regarding CPU usage. Top is working ok.

tvrtko@oxygene:~> cat /proc/stat
cpu 7452 0 2048 40426
cpu0 7452 0 2048 1844674407370950652
page 146199 106012
swap 1 0
intr 559574 499265 1318 0 0 0 0 1 673 2 0 0 3872 31224 0 23216 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
disk_io: (3,0):(23282,11382,292398,11900,212024)
ctxt 356071
btime 1070534465
processes 5491

#7 lore raw
Quoted 9 lines · Tvrtko A. Uršulin
Quoted 1 line · Con Kolivas

Updated my patchset.

There is something wrong with /proc/stat which confuses xosview and ksysguard regarding CPU usage. Top is working ok.

tvrtko@oxygene:~> cat /proc/stat
cpu 7452 0 2048 40426
cpu0 7452 0 2048 1844674407370950652

You're right. This looks like a hangover from the 64bit jiffies or the variable Hz. It's not a critical bug so unless someone can debug it for me it can wait.

Con

#8 lore raw
Quoted 11 lines · Con Kolivas
Quoted 6 lines · Tvrtko A. Uršulin

There is something wrong with /proc/stat which confuses xosview and ksysguard regarding CPU usage. Top is working ok.

tvrtko@oxygene:~> cat /proc/stat
cpu 7452 0 2048 40426
cpu0 7452 0 2048 1844674407370950652

You're right. This looks like a hangover from the 64bit jiffies or the variable Hz. It's not a critical bug so unless someone can debug it for me it can wait.

Oops, yes. I made a mistake when merging variable HZ with 64bit jiffies. Please apply the following patch on top of that.
Could you please confirm this fixes things, and could you also check that btime stays constant with this patch? Thanks!

Warning: totally untested. Pretty much obvious, however.

Sorry,
Tim

fs/proc/proc_misc.c +2 -2
@@ -422,7 +422,7 @@
 			(unsigned long long) jiffies_64_to_clock_t(user),
 			(unsigned long long) jiffies_64_to_clock_t(nice),
 			(unsigned long long) jiffies_64_to_clock_t(system),
-			(unsigned long long) jiffies_64_to_clock_t(jif - user - nice - system));
+			(unsigned long long) jif - jiffies_64_to_clock_t(user + nice + system));
 	}
 	proc_sprintf(page, &off, &len,
 		"page %u %u\n"
@@ -460,7 +460,7 @@
 		}
 	}
 
-	do_div(jif, HZ);
+	do_div(jif, USER_HZ);
 	proc_sprintf(page, &off, &len,
 		"\nctxt %lu\n"
 		"btime %lu\n"
#9 lore raw
Quoted 1 line · Tim Schmielau

Warning: totally untested. Pretty much obvious, however.

oxygene:/usr/src/linux-2.4.23-ck1 # patch --dry-run <../patches/2.4.23-ck1-stat-fix.patch -p1 --global-reject=bla patching file fs/proc/proc_misc.c
Hunk #1 FAILED at 422.
Hunk #2 FAILED at 460.
2 out of 2 hunks FAILED -- saving rejects to file fs/proc/proc_misc.c.rej

oxygene:/usr/src/linux-2.4.23-ck1 # cat bla
*************** fs/proc/proc_misc.c
*** 422,428 ****
(unsigned long long) jiffies_64_to_clock_t(user), (unsigned long long) jiffies_64_to_clock_t(nice), (unsigned long long) jiffies_64_to_clock_t(system),
- (unsigned long long) jiffies_64_to_clock_t(jif - user
-

 nice - system));
         }
         proc_sprintf(page, &off, &len,
                 "page %u %u\n"

--- 422,428 ----
(unsigned long long) jiffies_64_to_clock_t(user), (unsigned long long) jiffies_64_to_clock_t(nice), (unsigned long long) jiffies_64_to_clock_t(system),
+ (unsigned long long) jif - jiffies_64_to_clock_t(user
+

 nice + system));
         }
         proc_sprintf(page, &off, &len,
                 "page %u %u\n"
*************** fs/proc/proc_misc.c
*** 460,466 ****
                 }
         }
-        do_div(jif, HZ);
         proc_sprintf(page, &off, &len,
                 "\nctxt %lu\n"
                 "btime %lu\n"
--- 460,466 ----
                 }
         }
+        do_div(jif, USER_HZ);
         proc_sprintf(page, &off, &len,
                 "\nctxt %lu\n"
                 "btime %lu\n"
Quoted 24 lines

Sorry,
Tim

--- linux-2.4.23-ck1/fs/proc/proc_misc.c 2003-12-04 14:15:59.000000000
+0100 +++ linux-2.4.23-ck1-fix/fs/proc/proc_misc.c 2003-12-04
14:20:07.000000000 +0100 @@ -422,7 +422,7 @@

fs/proc/proc_misc.c +0 -1
 			(unsigned long long) jiffies_64_to_clock_t(user),
 			(unsigned long long) jiffies_64_to_clock_t(nice),
 			(unsigned long long) jiffies_64_to_clock_t(system),
-			(unsigned long long) jiffies_64_to_clock_t(jif - user - nice -

system)); + (unsigned long long) jif - jiffies_64_to_clock_t(user + nice
+ system)); }

fs/proc/proc_misc.c +1 -1
 	proc_sprintf(page, &off, &len,
 		"page %u %u\n"
@@ -460,7 +460,7 @@
 		}
 	}
 
-	do_div(jif, HZ);
+	do_div(jif, USER_HZ);
 	proc_sprintf(page, &off, &len,
 		"\nctxt %lu\n"
 		"btime %lu\n"
#10 lore raw
Quoted 9 lines · Tvrtko A. Uršulin
Quoted 1 line · Tim Schmielau

Warning: totally untested. Pretty much obvious, however.

oxygene:/usr/src/linux-2.4.23-ck1 # patch --dry-run <../patches/2.4.23-ck1-stat-fix.patch -p1 --global-reject=bla patching file fs/proc/proc_misc.c
Hunk #1 FAILED at 422.
Hunk #2 FAILED at 460.
2 out of 2 hunks FAILED -- saving rejects to file fs/proc/proc_misc.c.rej

Probably just a whitespace problem. Try this one. Please tell me if it works; I just made this based on what Tim sent (thanks muchly by the way Tim!)

Con

fs/proc/proc_misc.c +2 -2
diff -Naurp linux-2.4.23-ck1/fs/proc/proc_misc.c linux-2.4.23-ck1-fix1/fs/proc/proc_misc.c
@@ -422,7 +422,7 @@ static int kstat_read_proc(char *page, c
 			(unsigned long long) jiffies_64_to_clock_t(user),
 			(unsigned long long) jiffies_64_to_clock_t(nice),
 			(unsigned long long) jiffies_64_to_clock_t(system),
-			(unsigned long long) jiffies_64_to_clock_t(jif - user - nice - system));
+			(unsigned long long) jif - jiffies_64_to_clock_t(user + nice + system));
 	}
 	proc_sprintf(page, &off, &len,
 		"page %u %u\n"
@@ -460,7 +460,7 @@ static int kstat_read_proc(char *page, c
 		}
 	}
 
-	do_div(jif, HZ);
+	do_div(jif, USER_HZ);
 	proc_sprintf(page, &off, &len,
 		"\nctxt %lu\n"
 		"btime %lu\n"
#11 lore raw
Quoted 3 lines

Probably just a whitespace problem. Try this one. Please tell me if it works; I just made this based on what Tim sent (thanks muchly by the way Tim!)

Patched correctly, seems to work fine:

cpu 1247 0 891 14492
cpu0 1247 0 891 14491

Thank you both :)

Signature · 3 lines
Usuario Linux #212057 - Maquinas Linux #98894, #130864 y #168988
Proyecto LONIX: http://lonix.sourceforge.net
Lagrimas en la Lluvia: http://www.lagrimasenlalluvia.com