mm-sources 2.6.7 a bit too fast?

Message

fazto · Post by **fazto** » Tue Jun 22, 2004 6:38 am

Hi, I just upgraded from mm-sources-2.6.7-rc3-mm2 to mm-sources-2.6.7-mm1. I did a "make oldconfig" and build & installed the new kernel & modules. I rebooted and was pleasantly surprises by the performance boost

. I have no idea what has changed since rc3, but my notebook feels a _lot_ faster.

There is a problem however. Everything is faster, including the wait-time after a key is pressed to go into repeat-mode. So, typing anything in the box becomes a challenge by itself. For instance typing the word "linux" on a commandline in konsole results in "linnnuuxxxx"

So, what has changed? I like the new performance, but how to get rid of the "too short" key repeate timeout?

int2str · Post by **int2str** » Tue Jun 22, 2004 6:43 am

It's a known bug. Andew Morton posted two patches to the linux kernel mailing list to fix the problem.

Here:

Code: Select all

diff -puN arch/i386/kernel/mpparse.c~double-clock-speed-fix arch/i386/kernel/mpparse.c
--- 25/arch/i386/kernel/mpparse.c~double-clock-speed-fix        2004-06-20 23:28:16.655299120 -0700
+++ 25-akpm/arch/i386/kernel/mpparse.c  2004-06-20 23:28:20.468719392 -0700
@@ -1017,7 +1017,6 @@ void __init mp_config_acpi_legacy_irqs (
 
                for (idx = 0; idx < mp_irq_entries; idx++)
                        if (mp_irqs[idx].mpc_srcbus == MP_ISA_BUS &&
-                               (mp_irqs[idx].mpc_dstapic == ioapic) &&
                                (mp_irqs[idx].mpc_srcbusirq == i ||
                                mp_irqs[idx].mpc_dstirq == i))
                                        break;
diff -puN arch/x86_64/kernel/mpparse.c~double-clock-speed-fix arch/x86_64/kernel/mpparse.c
--- 25/arch/x86_64/kernel/mpparse.c~double-clock-speed-fix      2004-06-20 23:28:16.672296536 -0700
+++ 25-akpm/arch/x86_64/kernel/mpparse.c        2004-06-20 23:28:20.469719240 -0700
@@ -861,7 +861,6 @@ void __init mp_config_acpi_legacy_irqs (
 
                for (idx = 0; idx < mp_irq_entries; idx++)
                        if (mp_irqs[idx].mpc_srcbus == MP_ISA_BUS &&
-                               (mp_irqs[idx].mpc_dstapic == ioapic) &&
                                (mp_irqs[idx].mpc_srcbusirq == i ||
                                mp_irqs[idx].mpc_dstirq == i))
                                        break;

and here...

Code: Select all

diff -puN include/linux/kernel.h~abs-fix-fix include/linux/kernel.h
--- 25/include/linux/kernel.h~abs-fix-fix       2004-06-21 01:42:24.283873616 -0700
+++ 25-akpm/include/linux/kernel.h      2004-06-21 01:43:08.150204920 -0700
@@ -55,7 +55,12 @@ void __might_sleep(char *file, int line)
 #endif
 
 #define abs(x) ({                              \
-               typeof(x) __x = (x);            \
+               int __x = (x);                  \
+               (__x < 0) ? -__x : __x;         \
+       })
+
+#define labs(x) ({                             \
+               long __x = (x);                 \
                (__x < 0) ? -__x : __x;         \
        })

fazto · Post by **fazto** » Tue Jun 22, 2004 7:14 am

OK, thanks int2str.

I guess I must wait for mm2

. Will this also remove the really fast feel I have with this kernel, and if so, are there kernel configs to get that re-enabled?

andyknownasabu · Post by **andyknownasabu** » Tue Jun 22, 2004 8:08 am

fazto wrote:OK, thanks int2str.

I guess I must wait for mm2 :( . Will this also remove the really fast feel I have with this kernel, and if so, are there kernel configs to get that re-enabled?

I experienced exactly the same thing when I booted the new kernel for the first time this morning. I just thought WOW and fastened my seatbelt! ;)
For example all the menues in KDE popup a lot faster and the responsiveness of the whole system is dramatically increased.
Does anyone know what has been changed in kernel 2.6.7-mm1 to result in such an improvement?

I would also like to know, if with -mm2 these are gone?

UPDATE:
After a first successful compile of 2.6.7-mm1 yesterday evening, applying the above patch results in the following error message:

Code: Select all

  CC      arch/i386/kernel/time.o
In file included from arch/i386/kernel/time.c:59:
include/asm-i386/mach-default/mach_time.h: In Funktion >>mach_set_rtc_mmss<<:
include/asm-i386/mach-default/mach_time.h:51: error: redeclaration of `__x'
include/asm-i386/mach-default/mach_time.h:51: error: `__x' previously declared here
include/asm-i386/mach-default/mach_time.h:51: Warnung: unused variable `__x'
include/asm-i386/mach-default/mach_time.h:55: error: redeclaration of `__x'
include/asm-i386/mach-default/mach_time.h:55: error: `__x' previously declared here
include/asm-i386/mach-default/mach_time.h:55: Warnung: unused variable `__x'
make[1]: *** [arch/i386/kernel/time.o] Fehler 1
make: *** [arch/i386/kernel] Fehler 2

() · Post by () » Tue Jun 22, 2004 2:10 pm

The compile error is obviously the result of an erroneusly applied patch. Ie. the braces around the definition of abs were left out.

BlinkEye · Post by **BlinkEye** » Tue Jun 22, 2004 2:25 pm

i didn't have any problem at all. the kernel works right off the box. but thanks for the hint - the responsiveness of my system increased indeed to a known state (i don't know why but with more recent kernels i always had the feeling that if doing some things simultaneously it began to lag) - but it's all gone now. ah, that feels good!

elpollodiablo · Post by **elpollodiablo** » Tue Jun 22, 2004 4:51 pm

BlinkEye wrote:i didn't have any problem at all. the kernel works right off the box. but thanks for the hint - the responsiveness of my system increased indeed to a known state (i don't know why but with more recent kernels i always had the feeling that if doing some things simultaneously it began to lag) - but it's all gone now. ah, that feels good!

i'm having serious problems with my keyboard.... infact it does not work at all.... have u experinced something like that?

PS: i didn't apply any patch yet...

lighty14 · Post by **lighty14** » Tue Jun 22, 2004 5:01 pm

Is there a love-sources out using these new mm patches?

BlinkEye · Post by **BlinkEye** » Tue Jun 22, 2004 5:52 pm

elpollodiablo wrote:i'm having serious problems with my keyboard.... infact it does not work at all.... have u experinced something like that?

PS: i didn't apply any patch yet...

nope, i don't have any strange behavour at all and haven't applied any patch neither.
we're talking about this version of mm-sources, aren't we (i'm just asking as they're changing almost daily now - gentoo-dev-sources too):

Code: Select all

*  sys-kernel/mm-sources [ Masked ]
      Latest version available: 2.6.7-r1

besides:

Code: Select all

*  sys-kernel/gentoo-dev-sources
      Latest version available: 2.6.7-r3
      Latest version installed: 2.6.7-r3

is out too. maybe you wanna give a try. i'd had them installed - but i love mm-sources

elpollodiablo · Post by **elpollodiablo** » Tue Jun 22, 2004 6:04 pm

i'm talking about this:

Code: Select all

  sys-kernel/mm-sources [ Masked ]
      Latest version available: 2.6.7-r1

[/code]

no keyboard, my clock is running faster than it should, actually using gentoo-dev-sources....

fazto · Post by **fazto** » Tue Jun 22, 2004 6:07 pm

Yep, my clock is also running too fast, and I'm on mm-sources 2.6.7-mm1.

BlinkEye · Post by **BlinkEye** » Tue Jun 22, 2004 6:14 pm

well, it's a really fast kernel

no, i'm sorry you're experiencing problems. i can't help you but wanted to let you know that others don't have this problem.

sam_i_am · Post by **sam_i_am** » Tue Jun 22, 2004 6:46 pm

fazto wrote:Yep, my clock is also running too fast, and I'm on mm-sources 2.6.7-mm1.

Sammeeee heeereee. EEEvveeeeerything is ffffaaasssttttt

I eevveeennnn haaaaaave to ttype fffffffffaaaaaaaasssttttt tto ggget it rrright. AAaas youu cccannnn sssseeee, II don''t haaaaaave it maaassstered yettt.

Sam

[Edit]
Whew, after the patch was applied, everything is back to normal. I was a bit nervous as I couldn't go back to the old kernel (2.6.5-mm4) as X server would lock up with an error about a timer value not being what it expects. It seems to persist even through a power-off.

ank · Post by **ank** » Tue Jun 22, 2004 6:59 pm

Same here. Clock running like every 20s=1min and xmms scroll song names faster than usual.

searcher · Post by **searcher** » Tue Jun 22, 2004 7:32 pm

Yep, everything is running fast, but compiling and startup times are much faster too! Too bad the clock is messed up, but anything that uses the RTC for it's timing needs (such as mplayer) is oké. Xine messes up big time though.

I wonder if this can be made into a patch and the problems worked out, this is a major increase in speed for me, strangely enough from 2.6.7-rc3-love1.

~searcher

int2str · Post by **int2str** » Tue Jun 22, 2004 10:47 pm

The patches I posted fix the clock running fast problem as well as they key repeat problem.

nesl247 · Post by **nesl247** » Tue Jun 22, 2004 11:02 pm

How do i get bootsplash? I cant live without it.. Ok i can i just like it..

squeegy · Post by **squeegy** » Wed Jun 23, 2004 5:15 am

iotc247 wrote:How do i get bootsplash? I cant live without it.. Ok i can i just like it..

http://www.bootsplash.de/files/bootspla ... 7-mm1.diff

There's the bootsplash patch for mm1.

nesl247 · Post by **nesl247** » Wed Jun 23, 2004 1:51 pm

Does that patch make the mm go slower? Cause now it doesnt feel as fast..

[EDIT]Talking about the one for the keyboard[/EDIT]

huw · Post by **huw** » Wed Jun 23, 2004 3:03 pm

hmm

My clock runs too fast as well... Keyboard repeat is fine though.

Vim segfaults too!

ank · Post by **ank** » Wed Jun 23, 2004 3:41 pm

I couldn't patch kernel.
Here is mpparse.c.rej

***************
*** 1017,1023 ****

for (idx = 0; idx < mp_irq_entries; idx++)
if (mp_irqs[idx].mpc_srcbus == MP_ISA_BUS &&
- (mp_irqs[idx].mpc_dstapic == ioapic) &&
(mp_irqs[idx].mpc_srcbusirq == i ||
mp_irqs[idx].mpc_dstirq == i))
break;
--- 1017,1022 ----

for (idx = 0; idx < mp_irq_entries; idx++)
if (mp_irqs[idx].mpc_srcbus == MP_ISA_BUS &&
(mp_irqs[idx].mpc_srcbusirq == i ||
mp_irqs[idx].mpc_dstirq == i))
break;

ayuer · Post by **ayuer** » Wed Jun 23, 2004 4:14 pm

searcher wrote:Yep, everything is running fast, but compiling and startup times are much faster too! Too bad the clock is messed up, but anything that uses the RTC for it's timing needs (such as mplayer) is oké. Xine messes up big time though.

I wonder if this can be made into a patch and the problems worked out, this is a major increase in speed for me, strangely enough from 2.6.7-rc3-love1.

~searcher

Where could I find the love-sources ebuild, I cann't find anything on steel300's site, if you know, could you post a link, thanks

luqas · Post by **luqas** » Wed Jun 23, 2004 4:22 pm

I don't think the love-sources are ready yet. Anyway goto #love-sources on freenode in IRC or http://www.love-sources.org.

DZello · Post by **DZello** » Wed Jun 23, 2004 5:15 pm

ank wrote:I couldn't patch kernel.
Here is mpparse.c.rej

***************
*** 1017,1023 ****

for (idx = 0; idx < mp_irq_entries; idx++)
if (mp_irqs[idx].mpc_srcbus == MP_ISA_BUS &&
- (mp_irqs[idx].mpc_dstapic == ioapic) &&
(mp_irqs[idx].mpc_srcbusirq == i ||
mp_irqs[idx].mpc_dstirq == i))
break;
--- 1017,1022 ----

for (idx = 0; idx < mp_irq_entries; idx++)
if (mp_irqs[idx].mpc_srcbus == MP_ISA_BUS &&
(mp_irqs[idx].mpc_srcbusirq == i ||
mp_irqs[idx].mpc_dstirq == i))
break;

I have the same problem here.

Artherio · Post by **Artherio** » Wed Jun 23, 2004 6:09 pm

Sorryy for being a nooob,,, bbut howw ddOOO Ii applllly thhe patch? This is kind of annyoiing.