[REGRESSION] 2.6.25 on parisc - CPU 1 is stuck

[REGRESSION] 2.6.25 on parisc - CPU 1 is stuck

[Domenico Andreoli]

Hi,

  as soon as I booted my dual PA8600 with 2.6.25 I got 1 CPU less. The
only other kernel I successfully used on that box is Debian's 2.6.22-3.

Debian never built 2.6.23 for parisc, 2.6.24 was still hanging last
time I checked it, 2.6.25 did not even build until recent -rc. So I am
not in the best position to bisect.

Here is the relevant dmesg.

Cheers,
Domenico

Linux version 2.6.25-02519-g3925e6f (cavok@ska) (gcc version 4.1.3 20080308 (prerelease) (Debian 4.1.2-21)) #4 SMP Sun Apr 20 04:53:48
FP[0] enabled: Rev 1 Model 16
The 64-bit Kernel has started...
console [ttyB0] enabled
Initialized PDC Console for debugging.
Determining PDC firmware type: System Map.
model 00005d10 00000491 00000000 00000002 77b406fc 100000f0 00000008 000000b2 000000b2
vers  00000300
CPUID vers 17 rev 10 (0x0000022a)
capabilities 0x3
model 9000/785/J5600
Total Memory: 3840 MB
LCD display at fffffff0f05d0008,fffffff0f05d0000 registered
SMP: bootstrap CPU ID is 0
Built 1 zonelists in Zone order, mobility grouping on.  Total pages: 969600
Kernel command line: root=/dev/sdb5 panic=60 HOME=/ console=ttyS0 TERM=vt102 palo_kernel=2/vmlinux
PID hash table entries: 4096 (order: 12, 32768 bytes)
Console: colour dummy device 160x64
Dentry cache hash table entries: 524288 (order: 10, 4194304 bytes)
Inode-cache hash table entries: 262144 (order: 9, 2097152 bytes)
Memory: 3858944k/3932160k available (2916k kernel code, 72892k reserved, 1283k data, 240k init)
virtual kernel memory layout:
    vmalloc : 0x0000000000008000 - 0x000000003f000000   (1007 MB)
    memory  : 0x0000000040000000 - 0x0000000130000000   (3840 MB)
      .init : 0x00000000405b0000 - 0x00000000405ec000   ( 240 kB)
      .data : 0x00000000403d9160 - 0x000000004051a000   (1283 kB)
      .text : 0x0000000040100000 - 0x00000000403d9160   (2916 kB)
SLUB: Genslabs=12, HWalign=64, Order=0-1, MinObjects=4, CPUs=2, Nodes=1
Security Framework initialized
Capability LSM initialized
Mount-cache hash table entries: 256
Brought up 1 CPUs
net_namespace: 1168 bytes
NET: Registered protocol family 16
EISA bus registered
Searching for devices...
Found devices:
1. Astro BC Runway Port at 0xfffffffffed00000 [10] { 12, 0x0, 0x582, 0x0000b }
2. Elroy PCI Bridge at 0xfffffffffed30000 [10/0] { 13, 0x0, 0x782, 0x0000a }
3. Elroy PCI Bridge at 0xfffffffffed32000 [10/1] { 13, 0x0, 0x782, 0x0000a }
4. Elroy PCI Bridge at 0xfffffffffed34000 [10/2] { 13, 0x0, 0x782, 0x0000a }
5. Elroy PCI Bridge at 0xfffffffffed38000 [10/4] { 13, 0x0, 0x782, 0x0000a }
6. Elroy PCI Bridge at 0xfffffffffed3c000 [10/6] { 13, 0x0, 0x782, 0x0000a }
7. Forte W+ 2w at 0xfffffffffffa0000 [32] { 0, 0x0, 0x5d1, 0x00004 }
8. Forte W+ 2w at 0xfffffffffffa2000 [34] { 0, 0x0, 0x5d1, 0x00004 }
9. Memory at 0xfffffffffed10200 [49] { 1, 0x0, 0x09e, 0x00009 }
Enabling regular chassis codes support v0.05
Releasing cpu 1 now, hpa=fffffffffffa2000
SMP: CPU:1 is stuck.
CPU(s): 2 x PA8600 (PCX-W+) at 552.000000 MHz
Setting cache flush threshold to f84c0 (1 CPUs online)
SBA found Astro 2.1 at 0xfffffffffed00000
Elroy version TR2.1 (0x2) found at 0xfffffffffed30000
PCI: Enabled native mode for NS87415 (pif=0x8f)
Elroy version TR2.1 (0x2) found at 0xfffffffffed32000
Elroy version TR2.1 (0x2) found at 0xfffffffffed34000
Elroy version TR2.1 (0x2) found at 0xfffffffffed38000
Elroy version TR2.1 (0x2) found at 0xfffffffffed3c000
iosapic: hpa not registered for 0000:04:07.0
powersw: Soft power switch at 0xfffffff0f0400804 enabled.
...

-----[ Domenico Andreoli, aka cavok
 --[ http://www.dandreoli.com/gpgkey.asc
   ---[ 3A0F 2F80 F79C 678A 8936  4FEE 0677 9033 A20E BC50

Re: [REGRESSION] 2.6.25 on parisc - CPU 1 is stuck

[Grant Grundler]

On Mon, Apr 21, 2008 at 01:18:18AM +0200, Domenico Andreoli wrote:
> Hi,
> 
>   as soon as I booted my dual PA8600 with 2.6.25 I got 1 CPU less. The
> only other kernel I successfully used on that box is Debian's 2.6.22-3.

Sorry, I can't reproduce this on my j6000.

But there are some differences:
1) I'm using gcc 4.2
2) I'm testing 32-bit kernels.
3) j6k has 2GB of RAM (ie no RAM mapped above 4GB).
4) config file might be different in additional ways.

I'll kick off a 64-bit build and test that as well.


> Debian never built 2.6.23 for parisc, 2.6.24 was still hanging last
> time I checked it, 2.6.25 did not even build until recent -rc. So I am
> not in the best position to bisect.

Given 32-bit works, we might be able to figure this out w/o bisect.

...
> Linux version 2.6.25-02519-g3925e6f (cavok@ska) (gcc version 4.1.3 20080308 (prerelease) (Debian 4.1.2-21)) #4 SMP Sun Apr 20 04:53:48
...
> model 9000/785/J5600
> Total Memory: 3840 MB

Is this really 4GB with IO Hole?
The memory behind the IO Hole should be visible at a phys addr above 4GB.

...
> Found devices:
> 1. Astro BC Runway Port at 0xfffffffffed00000 [10] { 12, 0x0, 0x582, 0x0000b }
> 2. Elroy PCI Bridge at 0xfffffffffed30000 [10/0] { 13, 0x0, 0x782, 0x0000a }
> 3. Elroy PCI Bridge at 0xfffffffffed32000 [10/1] { 13, 0x0, 0x782, 0x0000a }
> 4. Elroy PCI Bridge at 0xfffffffffed34000 [10/2] { 13, 0x0, 0x782, 0x0000a }
> 5. Elroy PCI Bridge at 0xfffffffffed38000 [10/4] { 13, 0x0, 0x782, 0x0000a }
> 6. Elroy PCI Bridge at 0xfffffffffed3c000 [10/6] { 13, 0x0, 0x782, 0x0000a }
> 7. Forte W+ 2w at 0xfffffffffffa0000 [32] { 0, 0x0, 0x5d1, 0x00004 }
> 8. Forte W+ 2w at 0xfffffffffffa2000 [34] { 0, 0x0, 0x5d1, 0x00004 }
> 9. Memory at 0xfffffffffed10200 [49] { 1, 0x0, 0x09e, 0x00009 }
> Enabling regular chassis codes support v0.05
> Releasing cpu 1 now, hpa=fffffffffffa2000
> SMP: CPU:1 is stuck.

BTW, I'm not sure this is a kernel problem. Do older kernels still work?

> CPU(s): 2 x PA8600 (PCX-W+) at 552.000000 MHz
> Setting cache flush threshold to f84c0 (1 CPUs online)
...

FWIW, here's the equivalent output for the kernel I just booted on j6k:

Linux version 2.6.25gcc42 (grundler@j6k) (gcc version 4.2.3 (Debian 4.2.3-3)) #1 SMP Sat Apr 19 16:56:57 PDT 2008
...
model 9000/785/J6000
Total Memory: 2048 MB
...
Found devices:
1. Astro BC Runway Port at 0xfed00000 [10] { 12, 0x0, 0x582, 0x0000b }
2. Elroy PCI Bridge at 0xfed30000 [10/0] { 13, 0x0, 0x782, 0x0000a }
3. Elroy PCI Bridge at 0xfed34000 [10/2] { 13, 0x0, 0x782, 0x0000a }
4. Elroy PCI Bridge at 0xfed38000 [10/4] { 13, 0x0, 0x782, 0x0000a }
5. Elroy PCI Bridge at 0xfed3c000 [10/6] { 13, 0x0, 0x782, 0x0000a }
6. Duet W+ at 0xfffa0000 [32] { 0, 0x0, 0x5d4, 0x00004 }
7. Duet W+ at 0xfffa2000 [34] { 0, 0x0, 0x5d4, 0x00004 }
8. Memory at 0xfed10200 [49] { 1, 0x0, 0x00a, 0x00009 }
Enabling regular chassis codes support v0.05
Releasing cpu 1 now, hpa=fffa2000
FP[1] enabled: Rev 1 Model 16
CPU(s): 2 x PA8600 (PCX-W+) at 552.000000 MHz
Whole cache flush 240696 cycles, flushing 3936256 bytes 584013 cycles
Setting cache flush threshold to 1140 (2 CPUs online)
...

hth,
grant

Re: [REGRESSION] 2.6.25 on parisc - CPU 1 is stuck

[John David Anglin]

> > Releasing cpu 1 now, hpa=fffffffffffa2000
> > SMP: CPU:1 is stuck.
> 
> BTW, I'm not sure this is a kernel problem. Do older kernels still work?

I'm not sure this is related but a few months ago I noticed that only
two of the four cpus in the rp3440 that I have were running.  I reenabled
the two cpus that were not running, but I had to do it a couple of more
times before the two cpus remained enabled across reboots.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc-cnrc.gc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6602)

Re: [REGRESSION] 2.6.25 on parisc - CPU 1 is stuck

[Domenico Andreoli]

On Sun, Apr 20, 2008 at 11:00:29PM -0600, Grant Grundler wrote:
> On Mon, Apr 21, 2008 at 01:18:18AM +0200, Domenico Andreoli wrote:
> > 
> >   as soon as I booted my dual PA8600 with 2.6.25 I got 1 CPU less. The
> > only other kernel I successfully used on that box is Debian's 2.6.22-3.
> 
> Sorry, I can't reproduce this on my j6000.
>
> But there are some differences:
> 1) I'm using gcc 4.2

how do you build with gcc 4.2? if i try with make -j2 CC=gcc-4.2 build
fails almost immediately (yes, I have package gcc-4.2).

> 2) I'm testing 32-bit kernels.

indeeed disabling 64 bit support brings the cpu back

> ...
> > model 9000/785/J5600
> > Total Memory: 3840 MB
> 
> Is this really 4GB with IO Hole?
> The memory behind the IO Hole should be visible at a phys addr above 4GB.

really don't know, please read the diff between the two boots. thank you.

--- dmesg.txt	2008-04-21 11:19:16.000000000 +0200
+++ dmesg2.txt	2008-04-21 15:27:46.000000000 +0200
@@ -1,6 +1,7 @@
-Linux version 2.6.22-3-parisc64-smp (Debian 2.6.22-6) (maks@debian.org) (gcc version 4.1.3 20071019 (prerelease) (Debian 4.1.2-17)) #2 SMP Mon Nov 12 21:04:20 CET 2007
+Linux version 2.6.25 (cavok@ska) (gcc version 4.1.3 20080308 (prerelease) (Debian 4.1.2-21)) #8 SMP Mon Apr 21 12:25:50 CES8
 FP[0] enabled: Rev 1 Model 16
 The 64-bit Kernel has started...
+console [ttyB0] enabled
 Initialized PDC Console for debugging.
 Determining PDC firmware type: System Map.
 model 00005d10 00000491 00000000 00000002 77b406fc 100000f0 00000008 000000b2 000000b2
@@ -8,41 +9,28 @@
 CPUID vers 17 rev 10 (0x0000022a)
 capabilities 0x3
 model 9000/785/J5600
-Memory Ranges:
- 0) Start 0x0000000000000000 End 0x00000000efffffff Size   3840 MB
- 1) Start 0x00000010f0000000 End 0x00000010ffffffff Size    256 MB
-Total Memory: 4096 MB
-initrd: 7fa75000-7ffee77e
-initrd: reserving 3fa75000-3ffee77e (mem_max 100000000)
-On node 0 totalpages: 983040
-  Normal zone: 13440 pages used for memmap
-  Normal zone: 0 pages reserved
-  Normal zone: 969600 pages, LIFO batch:31
-On node 1 totalpages: 65536
-  Normal zone: 896 pages used for memmap
-  Normal zone: 0 pages reserved
-  Normal zone: 64640 pages, LIFO batch:15
+Total Memory: 3840 MB
 LCD display at fffffff0f05d0008,fffffff0f05d0000 registered
 SMP: bootstrap CPU ID is 0
-Built 2 zonelists.  Total pages: 1034240
-Kernel command line: root=/dev/sdb5 panic=60 HOME=/ console=ttyS0 TERM=vt102 palo_kernel=2/vmlinux.failsafe
+Built 1 zonelists in Zone order, mobility grouping on.  Total pages: 969600
+Kernel command line: root=/dev/sdb5 panic=60 HOME=/ console=ttyS0 TERM=vt102 palo_kernel=2/vmlinux
 PID hash table entries: 4096 (order: 12, 32768 bytes)
 Console: colour dummy device 160x64
 Dentry cache hash table entries: 524288 (order: 10, 4194304 bytes)
 Inode-cache hash table entries: 262144 (order: 9, 2097152 bytes)
-Memory: 4110336k/4194304k available (3041k kernel code, 83172k reserved, 1302k data, 296k init)
+Memory: 3858944k/3932160k available (2925k kernel code, 72904k reserved, 1278k data, 236k init)
 virtual kernel memory layout:
     vmalloc : 0x0000000000008000 - 0x000000003f000000   (1007 MB)
-    memory  : 0x0000000040000000 - 0x0000001140000000   (69632 MB)
-      .init : 0x0000000040608000 - 0x0000000040652000   ( 296 kB)
-      .data : 0x00000000403f8500 - 0x000000004053e000   (1302 kB)
-      .text : 0x0000000040100000 - 0x00000000403f8500   (3041 kB)
-Calibrating delay loop... 1101.82 BogoMIPS (lpj=2203648)
-Security Framework v1.0.0 initialized
-SELinux:  Disabled at boot.
+    memory  : 0x0000000040000000 - 0x0000000130000000   (3840 MB)
+      .init : 0x00000000405b4000 - 0x00000000405ef000   ( 236 kB)
+      .data : 0x00000000403db720 - 0x000000004051b000   (1278 kB)
+      .text : 0x0000000040100000 - 0x00000000403db720   (2925 kB)
+SLUB: Genslabs=12, HWalign=64, Order=0-1, MinObjects=4, CPUs=2, Nodes=1
+Security Framework initialized
 Capability LSM initialized
 Mount-cache hash table entries: 256
 Brought up 1 CPUs
+net_namespace: 1008 bytes
 NET: Registered protocol family 16
 EISA bus registered
 Searching for devices...
@@ -58,11 +46,9 @@
 9. Memory at 0xfffffffffed10200 [49] { 1, 0x0, 0x09e, 0x00009 }
 Enabling regular chassis codes support v0.05
 Releasing cpu 1 now, hpa=fffffffffffa2000
-FP[1] enabled: Rev 1 Model 16
-migration_cost=4000
+SMP: CPU:1 is stuck.
 CPU(s): 2 x PA8600 (PCX-W+) at 552.000000 MHz
-Whole cache flush 177618 cycles, flushing 5578752 bytes 826963 cycles
-Setting cache flush threshold to 100000 (2 CPUs online)
+Setting cache flush threshold to fc880 (1 CPUs online)
 SBA found Astro 2.1 at 0xfffffffffed00000
 Elroy version TR2.1 (0x2) found at 0xfffffffffed30000
 PCI: Enabled native mode for NS87415 (pif=0x8f)

-----[ Domenico Andreoli, aka cavok
 --[ http://www.dandreoli.com/gpgkey.asc
   ---[ 3A0F 2F80 F79C 678A 8936  4FEE 0677 9033 A20E BC50

Re: [REGRESSION] 2.6.25 on parisc - CPU 1 is stuck

[Grant Grundler]

On Mon, Apr 21, 2008 at 03:31:13PM +0200, Domenico Andreoli wrote:
> On Sun, Apr 20, 2008 at 11:00:29PM -0600, Grant Grundler wrote:
> > On Mon, Apr 21, 2008 at 01:18:18AM +0200, Domenico Andreoli wrote:
> > > 
> > >   as soon as I booted my dual PA8600 with 2.6.25 I got 1 CPU less. The
> > > only other kernel I successfully used on that box is Debian's 2.6.22-3.
> > 
> > Sorry, I can't reproduce this on my j6000.
> >
> > But there are some differences:
> > 1) I'm using gcc 4.2
> 
> how do you build with gcc 4.2? if i try with make -j2 CC=gcc-4.2 build
> fails almost immediately (yes, I have package gcc-4.2).

You probably need to change the .config so CONFIG_64BIT is not set.

_OR_ use the 64-bit compiler:
	make -j2 CC=hppa64-gnu-linux-gcc-4.2

> > 2) I'm testing 32-bit kernels.
> 
> indeeed disabling 64 bit support brings the cpu back

Ok. So it's not likely to be a HW problem.

> 
> > ...
> > > model 9000/785/J5600
> > > Total Memory: 3840 MB
> > 
> > Is this really 4GB with IO Hole?
> > The memory behind the IO Hole should be visible at a phys addr above 4GB.
> 
> really don't know, please read the diff between the two boots. thank you.

Sorry...that question was really targeting others on parisc-linux list.
The diff will some noise in it because 64-bit has to "F-extend" all
physical addresses to be 64-bit. (That's not quite the same as sign
extension because of some wonkiness in 32->64 bit phys address aliasing.)

thank,
grant

> --- dmesg.txt	2008-04-21 11:19:16.000000000 +0200
> +++ dmesg2.txt	2008-04-21 15:27:46.000000000 +0200
> @@ -1,6 +1,7 @@
> -Linux version 2.6.22-3-parisc64-smp (Debian 2.6.22-6) (maks@debian.org) (gcc version 4.1.3 20071019 (prerelease) (Debian 4.1.2-17)) #2 SMP Mon Nov 12 21:04:20 CET 2007
> +Linux version 2.6.25 (cavok@ska) (gcc version 4.1.3 20080308 (prerelease) (Debian 4.1.2-21)) #8 SMP Mon Apr 21 12:25:50 CES8
>  FP[0] enabled: Rev 1 Model 16
>  The 64-bit Kernel has started...
> +console [ttyB0] enabled
>  Initialized PDC Console for debugging.
>  Determining PDC firmware type: System Map.
>  model 00005d10 00000491 00000000 00000002 77b406fc 100000f0 00000008 000000b2 000000b2
> @@ -8,41 +9,28 @@
>  CPUID vers 17 rev 10 (0x0000022a)
>  capabilities 0x3
>  model 9000/785/J5600
> -Memory Ranges:
> - 0) Start 0x0000000000000000 End 0x00000000efffffff Size   3840 MB
> - 1) Start 0x00000010f0000000 End 0x00000010ffffffff Size    256 MB
> -Total Memory: 4096 MB
> -initrd: 7fa75000-7ffee77e
> -initrd: reserving 3fa75000-3ffee77e (mem_max 100000000)
> -On node 0 totalpages: 983040
> -  Normal zone: 13440 pages used for memmap
> -  Normal zone: 0 pages reserved
> -  Normal zone: 969600 pages, LIFO batch:31
> -On node 1 totalpages: 65536
> -  Normal zone: 896 pages used for memmap
> -  Normal zone: 0 pages reserved
> -  Normal zone: 64640 pages, LIFO batch:15
> +Total Memory: 3840 MB
>  LCD display at fffffff0f05d0008,fffffff0f05d0000 registered
>  SMP: bootstrap CPU ID is 0
> -Built 2 zonelists.  Total pages: 1034240
> -Kernel command line: root=/dev/sdb5 panic=60 HOME=/ console=ttyS0 TERM=vt102 palo_kernel=2/vmlinux.failsafe
> +Built 1 zonelists in Zone order, mobility grouping on.  Total pages: 969600
> +Kernel command line: root=/dev/sdb5 panic=60 HOME=/ console=ttyS0 TERM=vt102 palo_kernel=2/vmlinux
>  PID hash table entries: 4096 (order: 12, 32768 bytes)
>  Console: colour dummy device 160x64
>  Dentry cache hash table entries: 524288 (order: 10, 4194304 bytes)
>  Inode-cache hash table entries: 262144 (order: 9, 2097152 bytes)
> -Memory: 4110336k/4194304k available (3041k kernel code, 83172k reserved, 1302k data, 296k init)
> +Memory: 3858944k/3932160k available (2925k kernel code, 72904k reserved, 1278k data, 236k init)
>  virtual kernel memory layout:
>      vmalloc : 0x0000000000008000 - 0x000000003f000000   (1007 MB)
> -    memory  : 0x0000000040000000 - 0x0000001140000000   (69632 MB)
> -      .init : 0x0000000040608000 - 0x0000000040652000   ( 296 kB)
> -      .data : 0x00000000403f8500 - 0x000000004053e000   (1302 kB)
> -      .text : 0x0000000040100000 - 0x00000000403f8500   (3041 kB)
> -Calibrating delay loop... 1101.82 BogoMIPS (lpj=2203648)
> -Security Framework v1.0.0 initialized
> -SELinux:  Disabled at boot.
> +    memory  : 0x0000000040000000 - 0x0000000130000000   (3840 MB)
> +      .init : 0x00000000405b4000 - 0x00000000405ef000   ( 236 kB)
> +      .data : 0x00000000403db720 - 0x000000004051b000   (1278 kB)
> +      .text : 0x0000000040100000 - 0x00000000403db720   (2925 kB)
> +SLUB: Genslabs=12, HWalign=64, Order=0-1, MinObjects=4, CPUs=2, Nodes=1
> +Security Framework initialized
>  Capability LSM initialized
>  Mount-cache hash table entries: 256
>  Brought up 1 CPUs
> +net_namespace: 1008 bytes
>  NET: Registered protocol family 16
>  EISA bus registered
>  Searching for devices...
> @@ -58,11 +46,9 @@
>  9. Memory at 0xfffffffffed10200 [49] { 1, 0x0, 0x09e, 0x00009 }
>  Enabling regular chassis codes support v0.05
>  Releasing cpu 1 now, hpa=fffffffffffa2000
> -FP[1] enabled: Rev 1 Model 16
> -migration_cost=4000
> +SMP: CPU:1 is stuck.
>  CPU(s): 2 x PA8600 (PCX-W+) at 552.000000 MHz
> -Whole cache flush 177618 cycles, flushing 5578752 bytes 826963 cycles
> -Setting cache flush threshold to 100000 (2 CPUs online)
> +Setting cache flush threshold to fc880 (1 CPUs online)
>  SBA found Astro 2.1 at 0xfffffffffed00000
>  Elroy version TR2.1 (0x2) found at 0xfffffffffed30000
>  PCI: Enabled native mode for NS87415 (pif=0x8f)
> 
> -----[ Domenico Andreoli, aka cavok
>  --[ http://www.dandreoli.com/gpgkey.asc
>    ---[ 3A0F 2F80 F79C 678A 8936  4FEE 0677 9033 A20E BC50
> --
> To unsubscribe from this list: send the line "unsubscribe linux-parisc" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [REGRESSION] 2.6.25 on parisc - CPU 1 is stuck

[James Bottomley]

On Mon, 2008-04-21 at 01:18 +0200, Domenico Andreoli wrote: 
> as soon as I booted my dual PA8600 with 2.6.25 I got 1 CPU less. The
> only other kernel I successfully used on that box is Debian's 2.6.22-3.
> 
> Debian never built 2.6.23 for parisc, 2.6.24 was still hanging last
> time I checked it, 2.6.25 did not even build until recent -rc. So I am
> not in the best position to bisect.

Try this one; it's a corrected version of debian 2.6.24-1 for parisc
systems:

http://www.parisc-linux.org/~jejb/parisc_debs/linux-image-2.6.24-1-parisc64-smp_2.6.24-5+b1_hppa.deb

It's known to boot correctly on B, A and rp3440 systems, but we don't
have a J system to try it out on.

> Here is the relevant dmesg.
> Releasing cpu 1 now, hpa=fffffffffffa2000
> SMP: CPU:1 is stuck.
> CPU(s): 2 x PA8600 (PCX-W+) at 552.000000 MHz
> Setting cache flush threshold to f84c0 (1 CPUs online)

Unfortunately that doesn't tell us anything other than the start IPI
failed for some reason.  One thing to try with the J series is a 32 bit
kernel just in case: they're the only SMP systems that will boot 32
bits, just in case you could try a 32 bit smp kernel.

As Dave Anglin said, it could be a config problem, although when I have
CPUs deconfigured by firmware, they usually don't show up in the
inventory.

James

Re: [REGRESSION] 2.6.25 on parisc - CPU 1 is stuck

[Domenico Andreoli]

On 4/21/08, James Bottomley <James.Bottomley@hansenpartnership.com> wrote:
> On Mon, 2008-04-21 at 01:18 +0200, Domenico Andreoli wrote:
>  > as soon as I booted my dual PA8600 with 2.6.25 I got 1 CPU less. The
>  > only other kernel I successfully used on that box is Debian's 2.6.22-3.
>  >
>  > Debian never built 2.6.23 for parisc, 2.6.24 was still hanging last
>  > time I checked it, 2.6.25 did not even build until recent -rc. So I am
>  > not in the best position to bisect.
>
> Try this one; it's a corrected version of debian 2.6.24-1 for parisc
>  systems:
>
>  http://www.parisc-linux.org/~jejb/parisc_debs/linux-image-2.6.24-1-parisc64-smp_2.6.24-5+b1_hppa.deb
>
>  It's known to boot correctly on B, A and rp3440 systems, but we don't
>  have a J system to try it out on.

it does not work :( it stops here:

Linux version 2.6.24-1-parisc64-smp (Debian 2.6.24-5+b1)
(jejb@parisc-linux.org) (gcc version 4.1.3 20080114 (prerelease)
(Debian 4.1.2-19)) #2 SMP W8
FP[0] enabled: Rev 1 Model 16
The 64-bit Kernel has started...
console [ttyB0] enabled
Initialized PDC Console for debugging.
Determining PDC firmware type: System Map.
model 00005d10 00000491 00000000 00000002 77b406fc 100000f0 00000008
000000b2 000000b2
vers  00000300
CPUID vers 17 rev 10 (0x0000022a)
capabilities 0x3
model 9000/785/J5600
Memory Ranges:
 0) Start 0x0000000000000000 End 0x00000000efffffff Size   3840 MB
 1) Start 0x00000010f0000000 End 0x00000010ffffffff Size    256 MB
Total Memory: 4096 MB
initrd: 7fa06000-7ffee68b
initrd: reserving 3fa06000-3ffee68b (mem_max 100000000)
LCD display at fffffff0f05d0008,fffffff0f05d0000 registered
SMP: bootstrap CPU ID is 0
Built 2 zonelists in Zone order, mobility grouping on.  Total pages: 1034240
Kernel command line: root=/dev/sdb5 panic=60 HOME=/ console=ttyS0
TERM=vt102 palo_kernel=2/vmlinux
PID hash table entries: 4096 (order: 12, 32768 bytes)
Console: colour dummy device 160x64
Dentry cache hash table entries: 524288 (order: 10, 4194304 bytes)
Inode-cache hash table entries: 262144 (order: 9, 2097152 bytes)
Memory: 4109824k/4194304k available (3166k kernel code, 83792k
reserved, 1401k data, 296k init)
virtual kernel memory layout:
    vmalloc : 0x0000000000008000 - 0x000000003f000000   (1007 MB)
    memory  : 0x0000000040000000 - 0x0000001140000000   (69632 MB)
      .init : 0x000000004063c000 - 0x0000000040686000   ( 296 kB)
      .data : 0x0000000040417890 - 0x0000000040576000   (1401 kB)
      .text : 0x0000000040100000 - 0x0000000040417890   (3166 kB)
Security Framework initialized
SELinux:  Disabled at boot.
Capability LSM initialized
Mount-cache hash table entries: 256
Initializing cgroup subsys ns
Initializing cgroup subsys cpuacct
Brought up 1 CPUs
net_namespace: 120 bytes
NET: Registered protocol family 16
EISA bus registered
Searching for devices...
Found devices:
1. Astro BC Runway Port at 0xfffffffffed00000 [10] { 12, 0x0, 0x582, 0x0000b }
2. Elroy PCI Bridge at 0xfffffffffed30000 [10/0] { 13, 0x0, 0x782, 0x0000a }
3. Elroy PCI Bridge at 0xfffffffffed32000 [10/1] { 13, 0x0, 0x782, 0x0000a }
4. Elroy PCI Bridge at 0xfffffffffed34000 [10/2] { 13, 0x0, 0x782, 0x0000a }
5. Elroy PCI Bridge at 0xfffffffffed38000 [10/4] { 13, 0x0, 0x782, 0x0000a }
6. Elroy PCI Bridge at 0xfffffffffed3c000 [10/6] { 13, 0x0, 0x782, 0x0000a }
7. Forte W+ 2w at 0xfffffffffffa0000 [32] { 0, 0x0, 0x5d1, 0x00004 }
8. Forte W+ 2w at 0xfffffffffffa2000 [34] { 0, 0x0, 0x5d1, 0x00004 }
9. Memory at 0xfffffffffed10200 [49] { 1, 0x0, 0x09e, 0x00009 }
Enabling regular chassis codes support v0.05
Releasing cpu 1 now, hpa=fffffffffffa2000

-----[ Domenico Andreoli, aka cavok
 --[ http://www.dandreoli.com/gpgkey.asc
   ---[ 3A0F 2F80 F79C 678A 8936  4FEE 0677 9033 A20E BC50

Re: [REGRESSION] 2.6.25 on parisc - CPU 1 is stuck

[Grant Grundler]

On Mon, Apr 21, 2008 at 09:43:38AM -0500, James Bottomley wrote:
> On Mon, 2008-04-21 at 01:18 +0200, Domenico Andreoli wrote: 
> > as soon as I booted my dual PA8600 with 2.6.25 I got 1 CPU less. The
> > only other kernel I successfully used on that box is Debian's 2.6.22-3.
> > 
> > Debian never built 2.6.23 for parisc, 2.6.24 was still hanging last
> > time I checked it, 2.6.25 did not even build until recent -rc. So I am
> > not in the best position to bisect.
> 
> Try this one; it's a corrected version of debian 2.6.24-1 for parisc
> systems:
> 
> http://www.parisc-linux.org/~jejb/parisc_debs/linux-image-2.6.24-1-parisc64-smp_2.6.24-5+b1_hppa.deb

thanks! I can try that out tonight too.

> One thing to try with the J series is a 32 bit
> kernel just in case: they're the only SMP systems that will boot 32
> bits, just in case you could try a 32 bit smp kernel.

J-series are the only workstations that are SMP and boot 32-bit.
PA9000 K/D/R-class servers are SMP and can also boot 32-bit kernels.

> As Dave Anglin said, it could be a config problem, although when I have
> CPUs deconfigured by firmware, they usually don't show up in the
> inventory.

*nod* I don't expect CPUs to show up in inventory if they are
deconfigured in BCH. Maybe J-class firmware has this wrong? *shrug*
Something worth checking I guess.

thanks,
grant

Re: [REGRESSION] 2.6.25 on parisc - CPU 1 is stuck

[Kyle McMartin]

On Tue, Apr 22, 2008 at 06:44:40PM -0600, Grant Grundler wrote:
> 
> *nod* I don't expect CPUs to show up in inventory if they are
> deconfigured in BCH. Maybe J-class firmware has this wrong? *shrug*
> Something worth checking I guess.
> 

i booted the kernel from my rp3440 on my j6000. i also get this cpu1 is
stuck message. putting a panic in the SMP C-code entrypoint is never
hit.

most oddly, putting a hpmc-causing insn sequence in the entry point
doesn't trigger either, so the cpu isn't coming out of the rendezvous.

even more oddly, the stwa insn in the bring up code looks fine.

--kyle

Re: [REGRESSION] 2.6.25 on parisc - CPU 1 is stuck

[Domenico Andreoli]

On Tue, Apr 22, 2008 at 08:54:38PM -0400, Kyle McMartin wrote:
> On Tue, Apr 22, 2008 at 06:44:40PM -0600, Grant Grundler wrote:
> > 
> > *nod* I don't expect CPUs to show up in inventory if they are
> > deconfigured in BCH. Maybe J-class firmware has this wrong? *shrug*
> > Something worth checking I guess.
> > 
> 
> i booted the kernel from my rp3440 on my j6000. i also get this cpu1 is
> stuck message. putting a panic in the SMP C-code entrypoint is never
> hit.
> 
> most oddly, putting a hpmc-causing insn sequence in the entry point
> doesn't trigger either, so the cpu isn't coming out of the rendezvous.
> 
> even more oddly, the stwa insn in the bring up code looks fine.

still nothing changed with 2.6.26-rc1-00110-ga153063

cheers,
Domenico

-----[ Domenico Andreoli, aka cavok
 --[ http://www.dandreoli.com/gpgkey.asc
   ---[ 3A0F 2F80 F79C 678A 8936  4FEE 0677 9033 A20E BC50