From mboxrd@z Thu Jan 1 00:00:00 1970 From: Philippe Gerum In-Reply-To: <46BCA91A.1050209@domain.hid> References: <46BC4DCE.80003@domain.hid> <46BC5033.5040602@domain.hid> <1186761724.6006.6.camel@domain.hid> <46BCA91A.1050209@domain.hid> Content-Type: text/plain Date: Fri, 10 Aug 2007 20:11:50 +0200 Message-Id: <1186769510.6006.22.camel@domain.hid> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Sender: Philippe Gerum Subject: Re: [Xenomai-core] [2.6.22-BUG] inconsistent Linux IRQ mask Reply-To: rpm@xenomai.org List-Id: "Xenomai life and development \(bug reports, patches, discussions\)" List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Jan Kiszka Cc: xenomai-core On Fri, 2007-08-10 at 20:06 +0200, Jan Kiszka wrote: > Philippe Gerum wrote: > > On Fri, 2007-08-10 at 13:46 +0200, Jan Kiszka wrote: > >> Jan Kiszka wrote: > >>> Hi Philippe, > >>> > >>> this appears to be related to 2.6.22+ only: > >>> > >>> ------------[ cut here ]------------ > >>> kernel BUG at fs/buffer.c:1230! > >>> invalid opcode: 0000 [#1] > >>> PREEMPT > >>> Modules linked in: xeno_rtdm xeno_nucleus ip6t_LOG xt_TCPMSS nvram uinput snd_pcm_oss cifs snd_mixer_oss snd_seq snd_seq_device radeon drm wlan_tkip af_packet thinkpad_acpi button battery ac ip6t_REJECT xt_state iptable_nat nf_nat ip6table_mangle nf_conntrack_ipv4 nf_conntrack nfnetlink ip6table_filter ip6_tables ipv6 nls_iso8859_1 nls_cp437 vfat fat nls_utf8 ntfs dm_mod fuse wlan_scan_sta ath_rate_sample pcmcia snd_intel8x0 firmware_class snd_ac97_codec ac97_bus snd_pcm ath_pci nsc_ircc wlan ide_cd irda snd_timer e100 cdrom ath_hal(P) snd i2c_i801 uhci_hcd ehci_hcd mii yenta_socket rsrc_nonstatic soundcore usbcore snd_page_alloc i2c_core intel_agp crc_ccitt pcmcia_core agpgart parport_pc lp parport reiserfs edd fan piix ide_disk ide_core > >>> CPU: 0 > >>> EIP: 0060:[] Tainted: P VLI > >>> EFLAGS: 00210246 (2.6.22.2-adeos #7) > >>> EIP is at __find_get_block+0x2c/0x1e8 > >>> eax: 00000000 ebx: e3635dac ecx: 00000000 edx: 00002e4e > >>> esi: 00001000 edi: eeeb3080 ebp: e3635b30 esp: e3635af8 > >>> ds: 007b es: 007b fs: 0000 gs: 0033 ss: 0068 > >>> Process kio_file (pid: 7433, ti=e3634000 task=e6d780b0 task.ti=e3634000)<0> > >>> I-pipe domain Linux > >>> Stack: 00002e4e 00000000 c16e0718 eff36a90 e3634000 e3635b1c c0308f1e c05dd700 > >>> 00000000 e3635b2c c030bbfd e3635dac 00001000 e3635db4 e3635b98 c01881a4 > >>> 00001000 c05dd700 511c5a51 00000022 c014aade 00002e4e 00000000 eeeb3080 > >>> Call Trace: > >>> [] show_trace_log_lvl+0x1f/0x34 > >>> [] show_stack_log_lvl+0xa0/0xa8 > >>> [] show_registers+0x207/0x35c > >>> [] die+0x114/0x22a > >>> [] do_trap+0x8e/0xa7 > >>> [] do_invalid_op+0x8d/0x97 > >>> [] __ipipe_handle_exception+0xb1/0x11f > >>> [] error_code+0x6f/0x7c > >>> [] __getblk+0x25/0x29a > >>> [] search_by_key+0x97/0xe38 [reiserfs] > >>> [] search_by_entry_key+0x20/0x231 [reiserfs] > >>> [] reiserfs_find_entry+0x75/0x2d0 [reiserfs] > >>> [] reiserfs_lookup+0x6d/0x122 [reiserfs] > >>> ... > >>> > >>> I get this over 2.6.22.2 with latest trunk + ipipe-1.9-03, only > >>> modprobe'ing xeno_rtdm (so that the nucleus becomes active) and then > >>> logging into KDE. Same box, same Xenomai, but 2.6.20.15: no problem at > >>> all (also xeno-text runs fine here, BTW). When Xenomai is inactive, > >>> 2.6.22 makes no problems as well. Note that the tracer is compiled in, > >>> but is deactivated. Activating it (with the hope to get a panic > >>> backtrace) makes the issue disappear as well. > >> Ah, and then there was this warning once (same kernel that reliably BUGs): > >> > >> WARNING: at kernel/softirq.c:138 local_bh_enable() > >> [] show_trace_log_lvl+0x1f/0x34 > >> [] show_trace+0x17/0x19 > >> [] dump_stack+0x1a/0x1c > >> [] local_bh_enable+0x53/0xb3 > >> [] lock_sock_nested+0xbe/0xc6 > >> [] sock_fasync+0x68/0x125 > >> [] sock_close+0x2a/0x38 > >> [] __fput+0xb9/0x195 > >> [] fput+0x1b/0x1d > >> [] filp_close+0x59/0x61 > >> [] sys_close+0x73/0xab > >> [] sysenter_past_esp+0x6e/0x72 > >> > >> Probably a different symptom of the same bug. > >> > > > > Does this disappear with PREEMPT off? > > > > Nope. Good. > > Jan > -- Philippe.