pre-empt and smp in 2.5.37 - is it supposed to work?
I get a large screen full of hex addresses even before my framebuffer activates, so I wonder if breakage when using preempt and smp is a known issue in 2.5.37 or not?
.config:CONFIG_X86=yCONFIG_ISA=yCONFIG_UID16=yCONFIG_GENERIC_ISA_DMA=yCONFIG_EXPERIMENTAL=yCONFIG_NET=yCONFIG_SYSVIPC=yCONFIG_SYSCTL=yCONFIG_MODULES=yCONFIG_MODVERSIONS=yCONFIG_KMOD=yCONFIG_MPENTIUMIII=yCONFIG_X86_WP_WORKS_OK=yCONFIG_X86_INVLPG=yCONFIG_X86_CMPXCHG=yCONFIG_X86_XADD=yCONFIG_X86_BSWAP=yCONFIG_X86_POPAD_OK=yCONFIG_RWSEM_XCHGADD_ALGORITHM=yCONFIG_X86_TSC=yCONFIG_X86_GOOD_APIC=yCONFIG_X86_USE_PPRO_CHECKSUM=yCONFIG_SMP=yCONFIG_X86_MCE=yCONFIG_X86_MCE_NONFATAL=yCONFIG_X86_MCE_P4THERMAL=yCONFIG_HIGHMEM4G=yCONFIG_HIGHMEM=yCONFIG_MTRR=yCONFIG_HAVE_DEC_LOCK=yCONFIG_PM=yCONFIG_APM=yCONFIG_X86_IO_APIC=yCONFIG_X86_LOCAL_APIC=yCONFIG_X86_MPPARSE=yCONFIG_PCI=yCONFIG_PCI_GOANY=yCONFIG_PCI_BIOS=yCONFIG_PCI_DIRECT=yCONFIG_PCI_NAMES=yCONFIG_KCORE_ELF=yCONFIG_BINFMT_AOUT=yCONFIG_BINFMT_ELF=yCONFIG_BINFMT_MISC=mCONFIG_PARPORT=yCONFIG_PARPORT_PC=yCONFIG_PARPORT_PC_CML1=yCONFIG_PARPORT_1284=yCONFIG_BLK_DEV_FD=yCONFIG_BLK_DEV_LOOP=mCONFIG_IDE=yCONFIG_BLK_DEV_IDE=yCONFIG_BLK_DEV_IDEDISK=yCONFIG_BLK_DEV_IDECD=yCONFIG_BLK_DEV_IDESCSI=yCONFIG_BLK_DEV_IDEPCI=yCONFIG_IDEPCI_SHARE_IRQ=yCONFIG_BLK_DEV_IDEDMA_PCI=yCONFIG_IDEDMA_PCI_AUTO=yCONFIG_BLK_DEV_IDEDMA=yCONFIG_BLK_DEV_ADMA=yCONFIG_BLK_DEV_HPT366=yCONFIG_BLK_DEV_PDC202XX_OLD=yCONFIG_PDC202XX_BURST=yCONFIG_BLK_DEV_VIA82CXXX=yCONFIG_IDEDMA_AUTO=yCONFIG_BLK_DEV_PDC202XX=yCONFIG_BLK_DEV_IDE_MODES=yCONFIG_SCSI=yCONFIG_BLK_DEV_SD=yCONFIG_CHR_DEV_ST=yCONFIG_BLK_DEV_SR=yCONFIG_CHR_DEV_SG=yCONFIG_SCSI_MULTI_LUN=yCONFIG_SCSI_REPORT_LUNS=yCONFIG_SCSI_CONSTANTS=yCONFIG_SCSI_SYM53C8XX=yCONFIG_SCSI_NCR53C8XX_SYMBIOS_COMPAT=yCONFIG_MD=yCONFIG_BLK_DEV_MD=yCONFIG_MD_LINEAR=yCONFIG_MD_RAID0=yCONFIG_MD_RAID1=yCONFIG_MD_RAID5=yCONFIG_PACKET=yCONFIG_UNIX=yCONFIG_INET=yCONFIG_IP_MULTICAST=yCONFIG_IP_ADVANCED_ROUTER=yCONFIG_IP_ROUTE_VERBOSE=yCONFIG_IP_ROUTE_LARGE_TABLES=yCONFIG_SYN_COOKIES=yCONFIG_IPV6_SCTP__=yCONFIG_NETDEVICES=yCONFIG_DUMMY=mCONFIG_NET_ETHERNET=yCONFIG_NET_PCI=yCONFIG_8139TOO=mCONFIG_INPUT=yCONFIG_INPUT_MOUSEDEV=yCONFIG_INPUT_MOUSEDEV_PSAUX=yCONFIG_SOUND_GAMEPORT=yCONFIG_SERIO=yCONFIG_SERIO_I8042=yCONFIG_INPUT_KEYBOARD=yCONFIG_KEYBOARD_ATKBD=yCONFIG_INPUT_MOUSE=yCONFIG_MOUSE_PS2=yCONFIG_INPUT_MISC=yCONFIG_INPUT_PCSPKR=yCONFIG_VT=yCONFIG_VT_CONSOLE=yCONFIG_HW_CONSOLE=yCONFIG_SERIAL_8250=yCONFIG_SERIAL_CORE=yCONFIG_UNIX98_PTYS=yCONFIG_PRINTER=yCONFIG_I2C=mCONFIG_I2C_ALGOBIT=mCONFIG_I2C_CHARDEV=mCONFIG_WATCHDOG=yCONFIG_SOFT_WATCHDOG=yCONFIG_RTC=mCONFIG_AGP=mCONFIG_AGP_VIA=yCONFIG_DRM=yCONFIG_DRM_MGA=mCONFIG_RAW_DRIVER=yCONFIG_REISERFS_FS=yCONFIG_EXT3_FS=yCONFIG_JBD=yCONFIG_FAT_FS=mCONFIG_VFAT_FS=mCONFIG_TMPFS=yCONFIG_RAMFS=yCONFIG_ISO9660_FS=mCONFIG_JOLIET=yCONFIG_NTFS_FS=mCONFIG_PROC_FS=yCONFIG_DEVPTS_FS=yCONFIG_EXT2_FS=yCONFIG_UDF_FS=mCONFIG_MSDOS_PARTITION=yCONFIG_NLS=yCONFIG_NLS_CODEPAGE_437=yCONFIG_NLS_ISO8859_1=yCONFIG_VGA_CONSOLE=yCONFIG_VIDEO_SELECT=yCONFIG_FB=yCONFIG_DUMMY_CONSOLE=yCONFIG_VIDEO_SELECT=yCONFIG_FB_MATROX=yCONFIG_FB_MATROX_G450=yCONFIG_FB_MATROX_G100=yCONFIG_FB_MATROX_I2C=mCONFIG_FBCON_ADVANCED=yCONFIG_FBCON_CFB8=yCONFIG_FBCON_CFB16=yCONFIG_FBCON_CFB24=yCONFIG_FBCON_CFB32=yCONFIG_FBCON_FONTS=y
CONFIG_FONT_SUN12x22=yCONFIG_SOUND=yCONFIG_SND=yCONFIG_SND_SEQUENCER=yCONFIG_SND_OSSEMUL=yCONFIG_SND_MIXER_OSS=yCONFIG_SND_PCM_OSS=mCONFIG_SND_SEQUENCER_OSS=yCONFIG_SND_RTCTIMER=mCONFIG_SND_EMU10K1=yCONFIG_USB=yCONFIG_USB_DEVICEFS=yCONFIG_USB_UHCI_HCD_ALT=yCONFIG_USB_PRINTER=yCONFIG_USB_SCANNER=yCONFIG_DEBUG_KERNEL=yCONFIG_MAGIC_SYSRQ=yCONFIG_X86_EXTRA_IRQS=yCONFIG_X86_FIND_SMP_CONFIG=yCONFIG_SECURITY_CAPABILITIES=yCONFIG_X86_SMP=yCONFIG_X86_HT=yCONFIG_X86_BIOS_REBOOT=y
Jurriaan
Signature · 3 lines
If Big Brother is watching you, stare back, he doesn't like it. Shannon GNU/Linux 2.5.37 SMP/ReiserFS 2x1380 bogomips load av: 1.63 0.49 0.19
Quoted 3 lines · Jurriaan
I get a large screen full of hex addresses even before my framebuffer activates, so I wonder if breakage when using preempt and smp is a known issue in 2.5.37 or not?
You need this yet-to-be-merged patch. It should work fine with it.
It is just an overzealous debugging test..
Robert Love
kernel/sched.c +11 -2
@@ -940,8 +940,17 @@struct list_head *queue;int idx; -if (unlikely (in_atomic ())) -BUG (); +/* +* Test if we are atomic. Since do_exit() needs to call into +* schedule() atomically, we ignore that path for now. +* Otherwise, whine if we are scheduling when we should not be. +*/ +if (likely (current->state !=TASK_ZOMBIE )) { +if (unlikely (in_atomic ())) { +printk (KERN_ERR "bad: scheduling while atomic!\n" ); +dump_stack (); + } + }#if CONFIG_DEBUG_HIGHMEM check_highmem_ptes ();
Re: pre-empt and smp in 2.5.37 - is it supposed to work? [contains 2 oopses, one in set_cpus_allowed, one in md code]
From: Robert Love <[email protected]>
Date: Fri, Sep 20, 2002 at 04:09:22PM -0400
Quoted 26 lines · Jurriaan
Quoted 4 lines
I get a large screen full of hex addresses even before my framebuffer activates, so I wonder if breakage when using preempt and smp is a known issue in 2.5.37 or not?
You need this yet-to-be-merged patch. It should work fine with it.
It is just an overzealous debugging test..
Robert Love
diff -urN
linux-2.5.37/kernel/sched.clinux/kernel/sched.ckernel/sched.c +4
-2@@ -940,8 +940,17 @@struct list_head *queue;int idx; -if (unlikely (in_atomic ())) -BUG (); +/* +* Test if we are atomic. Since do_exit() needs to call into +* schedule() atomically, we ignore that path for now. +* Otherwise, whine if we are scheduling when we should not be.
kernel/sched.c line 956
Quoted 10 lines
kernel/sched.c +7
-0+*/ +if (likely (current->state !=TASK_ZOMBIE )) { +if (unlikely (in_atomic ())) { +printk (KERN_ERR "bad: scheduling while atomic!\n" ); +dump_stack (); + } + }#if CONFIG_DEBUG_HIGHMEM check_highmem_ptes ();
Well, then, perhaps you're interested in the dump?
I get two 'bad: scheduling while atomic' messages when booting:
ksymoops 2.4.6 on i686 2.5.37. Options used
-V (default)
-k /proc/ksyms (default)
-l /proc/modules (default)
-o /lib/modules/2.5.37/ (default)
-m /boot/System.map-2.5.37 (default)
Warning: You did not tell me where to find symbol information. I will assume that the log matches the kernel and modules that are running right now and I'll use the default options above for symbol resolution. If the current kernel and/or modules do not match the log, you can get more accurate output by telling me the kernel version and where to find map, modules, ksyms etc. ksymoops -h explains the options.
Error (regular_file): read_system_map stat /boot/System.map-2.5.37 failed
ksymoops: No such file or directory
cpu: 0, clocks: 100440, slice: 3043
cpu: 1, clocks: 100440, slice: 3043
CPU 1 IS NOW UP!c1b69f00 c01165f1 c02db020 c1b68000 c1b69f70 c1b69f78 00000000 00000000
00000000 00000000 00000000 00000000 c1b68000 c1b69f78 c0116c3c 00000000
c1b68000 c03e24a0 c1b68000 c1b69fa4 00000000 c1b75bc0 c01169fc 00000000
Call Trace: [<c01165f1>] [<c0116c3c>] [<c01169fc>] [<c01169fc>] [<c01180e6>]
[<c0118158>] [<c0118108>] [<c01054f1>]
c1b67f1c c01165f1 c02db020 c1b66000 c1b67f8c c1b67f94 00000000 00000000
00000000 00000000 00000000 00000000 c1b66000 c1b67f94 c0116c3c 00000000
c1b66000 c03e24a0 c1b66000 c1b67fc0 00000000 c1b762a0 c01169fc 00000000
Call Trace: [<c01165f1>] [<c0116c3c>] [<c01169fc>] [<c01169fc>] [<c01180e6>]
[<c012031d>] [<c01202cc>] [<c01054f1>] Warning (Oops_read): Code line not seen, dumping what data is available
Trace; c01165f1 <schedule+3d/404>
Trace; c0116c3c <wait_for_completion+9c/f8>
Trace; c01169fc <default_wake_function+0/80>
Trace; c01169fc <default_wake_function+0/80>
Trace; c01180e6 <set_cpus_allowed+13a/744>
Trace; c0118158 <set_cpus_allowed+1ac/744>
Trace; c0118108 <set_cpus_allowed+15c/744>
Trace; c01054f1 <enable_hlt+1c9/1d0>
Trace; c01165f1 <schedule+3d/404>
Trace; c0116c3c <wait_for_completion+9c/f8>
Trace; c01169fc <default_wake_function+0/80>
Trace; c01169fc <default_wake_function+0/80>
Trace; c01180e6 <set_cpus_allowed+13a/744>
Trace; c012031d <__run_task_queue+dd/168>
Trace; c01202cc <__run_task_queue+8c/168>
Trace; c01054f1 <enable_hlt+1c9/1d0>
8139too Fast Ethernet driver 0.9.26
Unable to handle kernel NULL pointer dereference at virtual address 0000000e c024b6b9 *pde = 00000000 Oops: 0000 CPU: 0 EIP: 0060:[<c024b6b9>] Not tainted
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010286
eax: 0000001e ebx: fffffffa ecx: c03675a8 edx: 00000292
esi: fffffffa edi: ffffffea ebp: 00002103 esp: f7737eec
ds: 0068 es: 0068 ss: 0068
Stack: 00002103 c024d562 fffffffa 00000931 00002103 00002103 00000000 c024dfd9
00002103 00000931 f7ca5e40 00002103 f77a3ee0 00000000 00000000 f778ba00
000061b0 00000001 00000000 00000006 00002103 00000000 00000000 00000000
Call Trace: [<c024d562>] [<c024dfd9>] [<c014a826>] [<c0152d39>] [<c01071eb>]
Code: 8b 43 14 85 c0 74 10 0f b7 50 10 b2 00 66 0f b6 40 10 09 c2 Quoted 4 lines
Quoted 4 lines
EIP;
c024b6b9<partition_name+799/aa0> <=====ecx;
c03675a8<abi_defhandler_libcso+1b0/2a4>
esp;f7737eec<ipv4_config+370cd784/3a2a98f8>
Trace; c024d562 <md_print_devices+1ba2/2d24>
Trace; c024dfd9 <md_print_devices+2619/2d24>
Trace; c014a826 <blkdev_put+2ca/318>
Trace; c0152d39 <kill_fasync+4cd/534>
Trace; c01071eb <__read_lock_failed+f0f/1c3c>
Code; c024b6b9 <partition_name+799/aa0>
00000000 <_EIP>:
Code; c024b6b9 <partition_name+799/aa0> <=====
0: 8b 43 14 mov 0x14(%ebx),%eax <=====
Code; c024b6bc <partition_name+79c/aa0>
3: 85 c0 test %eax,%eax
Code; c024b6be <partition_name+79e/aa0>
5: 74 10 je 17 <_EIP+0x17> c024b6d0 <partition_name+7b0/aa0>
Code; c024b6c0 <partition_name+7a0/aa0>
7: 0f b7 50 10 movzwl 0x10(%eax),%edx
Code; c024b6c4 <partition_name+7a4/aa0>
b: b2 00 mov $0x0,%dl
Code; c024b6c6 <partition_name+7a6/aa0>
d: 66 0f b6 40 10 movzbw 0x10(%eax),%ax
Code; c024b6cb <partition_name+7ab/aa0>
12: 09 c2 or %eax,%edx
2 warnings and 1 error issued. Results may not be reliable.
hde3 does no longer exist - I need to redo my configuration, but that shouldn't oops the kernel.
checking TSC synchronization across 2 CPUs: passed.
Starting migration thread for cpu 0
Bringing up 1
cpu: 1, clocks: 100440, slice: 3043 CPU1<T0:100432,T1:94336,D:10,S:3043,C:100440>
CPU 1 IS NOW UP!
Starting migration thread for cpu 1
bad: scheduling while atomic!c1b69f00 c01165f1 c02db020 c1b68000 c1b69f70 c1b69f78 00000000 00000000
00000000 00000000 00000000 00000000 c1b68000 c1b69f78 c0116c3c 00000000
c1b68000 c03e24a0 c1b68000 c1b69fa4 00000000 c1b75bc0 c01169fc 00000000
Call Trace: [<c01165f1>] [<c0116c3c>] [<c01169fc>] [<c01169fc>] [<c01180e6>]
[<c0118158>] [<c0118108>] [<c01054f1>]
bad: scheduling while atomic!
c1b67f1c c01165f1 c02db020 c1b66000 c1b67f8c c1b67f94 00000000 00000000
00000000 00000000 00000000 00000000 c1b66000 c1b67f94 c0116c3c 00000000
c1b66000 c03e24a0 c1b66000 c1b67fc0 00000000 c1b762a0 c01169fc 00000000
Call Trace: [<c01165f1>] [<c0116c3c>] [<c01169fc>] [<c01169fc>] [<c01180e6>]
[<c012031d>] [<c01202cc>] [<c01054f1>] CPUS done 4294967295
Linux NET4.0 for Linux 2.4
md: could not lock hde3.
md: could not import hde3!
Unable to handle kernel NULL pointer dereference at virtual address 0000000e
printing eip:
c024b6b9
*pde = 00000000
Oops: 0000
CPU: 0
EIP: 0060:[<c024b6b9>] Not tainted
EFLAGS: 00010286
eax: 0000001e ebx: fffffffa ecx: c03675a8 edx: 00000292
esi: fffffffa edi: ffffffea ebp: 00002103 esp: f7737eec
ds: 0068 es: 0068 ss: 0068
Process raidstart (pid: 89, threadinfo=f7736000 task=f7aa1520)
Stack: 00002103 c024d562 fffffffa 00000931 00002103 00002103 00000000 c024dfd9
00002103 00000931 f7ca5e40 00002103 f77a3ee0 00000000 00000000 f778ba00
000061b0 00000001 00000000 00000006 00002103 00000000 00000000 00000000
Call Trace: [<c024d562>] [<c024dfd9>] [<c014a826>] [<c0152d39>] [<c01071eb>]
Code: 8b 43 14 85 c0 74 10 0f b7 50 10 b2 00 66 0f b6 40 10 09 c2
<6>note: raidstart[89] exited with preempt_count 1found reiserfs format "3.6" with standard journal
Jurriaan
Signature · 3 lines
I am the special news bulletin that interrupts your favorite show Darkwing Duck GNU/Linux 2.5.37 SMP/ReiserFS 2x1380 bogomips load av: 0.15 0.21 0.20
Re: pre-empt and smp in 2.5.37 - is it supposed to work? [contains 2 oopses, one in set_cpus_allowed, one in md code]
Quoted 8 lines · Jurriaan
Trace;
c01165f1<schedule+3d/404>
Trace;c0116c3c<wait_for_completion+9c/f8>
Trace;c01169fc<default_wake_function+0/80>
Trace;c01169fc<default_wake_function+0/80>
Trace;c01180e6<set_cpus_allowed+13a/744>
Trace;c0118158<set_cpus_allowed+1ac/744>
Trace;c0118108<set_cpus_allowed+15c/744>
Trace;c01054f1<enable_hlt+1c9/1d0>
Quoted 8 lines
Trace;
c01165f1<schedule+3d/404>
Trace;c0116c3c<wait_for_completion+9c/f8>
Trace;c01169fc<default_wake_function+0/80>
Trace;c01169fc<default_wake_function+0/80>
Trace;c01180e6<set_cpus_allowed+13a/744>
Trace;c012031d<__run_task_queue+dd/168>
Trace;c01202cc<__run_task_queue+8c/168>
Trace;c01054f1<enable_hlt+1c9/1d0>
This is known, its due to set_cpus_allowed() sleeping while holding a preempt_disable(). It is harmless but something I need to fix.
Quoted 12 lines
8139too Fast Ethernet driver 0.9.26
Unable to handle kernel NULL pointer dereference at virtual address 0000000e c024b6b9 *pde = 00000000 Oops: 0000 CPU: 0 EIP: 0060:[<c024b6b9>] Not taintedUsing defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010286
eax:0000001eebx: fffffffa ecx:c03675a8edx: 00000292
esi: fffffffa edi: ffffffea ebp: 00002103 esp:f7737eec
ds: 0068 es: 0068 ss: 0068
Quoted 5 lines
Stack: 00002103 c024d562 fffffffa 00000931 00002103 00002103 00000000 c024dfd9
00002103 00000931 f7ca5e40 00002103 f77a3ee0 00000000 00000000 f778ba00
000061b0 00000001 00000000 00000006 00002103 00000000 00000000 00000000
Call Trace: [<c024d562>] [<c024dfd9>] [<c014a826>] [<c0152d39>] [<c01071eb>]
Code: 8b 43 14 85 c0 74 10 0f b7 50 10 b2 00 66 0f b6 40 10 09 c2 What is this? Do you normally see this?
Robert Love
Re: pre-empt and smp in 2.5.37 - is it supposed to work? [contains 2 oopses, one in set_cpus_allowed, one in md code]
From: Robert Love <[email protected]>
Date: Fri, Sep 20, 2002 at 05:29:26PM -0400
Quoted 20 lines · Jurriaan
Quoted 13 lines
8139too Fast Ethernet driver 0.9.26
Unable to handle kernel NULL pointer dereference at virtual address 0000000e c024b6b9 *pde = 00000000 Oops: 0000 CPU: 0 EIP: 0060:[<c024b6b9>] Not taintedUsing defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010286
eax:0000001eebx: fffffffa ecx:c03675a8edx: 00000292
esi: fffffffa edi: ffffffea ebp: 00002103 esp:f7737eec
ds: 0068 es: 0068 ss: 0068Quoted 5 lines
Stack: 00002103 c024d562 fffffffa 00000931 00002103 00002103 00000000 c024dfd9 00002103 00000931 f7ca5e40 00002103 f77a3ee0 00000000 00000000 f778ba00 000061b0 00000001 00000000 00000006 00002103 00000000 00000000 00000000 Call Trace: [<c024d562>] [<c024dfd9>] [<c014a826>] [<c0152d39>] [<c01071eb>] Code: 8b 43 14 85 c0 74 10 0f b7 50 10 b2 00 66 0f b6 40 10 09 c2
Quoted 2 lines
What is this? Do you normally see this?
As a matter of fact, no. The second boot of 2.5.37 didn't oops at all - perhaps I'll have better luck next reboot :-)
Jurriaan
Signature · 4 lines
Und wenn Ihr hoefflich zu mir seid Schau ich auch mal bei euch vorbei die Toten Hosen - Streichholzmann GNU/Linux 2.5.37 SMP/ReiserFS 2x1380 bogomips load av: 0.00 0.16 0.13