Bugzilla – Bug 115424
Hwinfo freezes in UML installation
Last modified: 2005-09-09 09:09:26 UTC
When I start UML installation using Yast module I get this output: Initialized stdio console driver Initializing software serial port version 1 Console initialized on /dev/ttyS0 ubda: unknown partition table Moving into tmpfs... done. created /dev tree (14551 inodes) >>> SUSE Linux installation program v1.9.13 (c) 1996-2005 SUSE Linux Products GmbH <<< line_ioctl: ttyS0: unknown ioctl: 0x4b45 Starting hardware detection... > sys.1: cpu
What does the kernel oops say?
BTW, if you can't get it at this point: use manual=1 or even manual=2 and run hwinfo from inst-sys or rescue system.
There is no kernel oops, hwinfo just stops. Content of the third console: Remount of / ok Console: /dev/ttyS0, serial line params "ttyS0" Please supply a config file error setting kbd mode: Invalid argument Window size: 80 x 24 CPU: 0, Memory: 232856 insmod /modules/af_packet.ko insmod /modules/nls_cp437.ko insmod /modules/nls_iso8859-1.ko insmod /modules/fat.ko insmod /modules/vfat.ko insmod /modules/isofs.ko Beginning hardware probing...
It's same even if I use 'probe=-cpu' boot parameter.
hwinfo doesn't just stop. :-) The idea is to run hwinfo through strace, that's why I asked you to use manual=2 to get past that point. BTW, the parameter is 'hwprobe=-sys'. If you use it, run 'hwprobe=+sys strace hwinfo --sys' after that to undo its effect. There is really nothing where it could hang at this point, except by reading some /proc data.
I used rescue system from Beta4plus, the end of strace output is: > bios.3: smp) = 79 open("/dev/mem", O_RDONLY) = 3 fstat64(3, {st_mode=S_IFCHR|0640, st_rdev=makedev(1, 1), ...}) = 0 mmap2(NULL, 4096, PROT_READ, MAP_PRIVATE, 3, 0x9f) = 0x40018000 munmap(0x40018000, 4096) = 0 close(3) = 0 > bios.5: 32) = 78 > sys.1: cpu) = 78 open("/proc/cpuinfo", O_RDONLY) = 3 fstat64(3, {st_mode=S_IFREG|0444, st_size=0, ...}) = 0 mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x40018000 read(3, "processor\t: 0\nvendor_id\t: User M"..., 1024) = 166 read(3, "", 1024) = 0 close(3) = 0 munmap(0x40018000, 4096) = 0 clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0x40528cc8) = 1359 waitpid(1359, [1]+ Stopped hwprobe=+sys strace hwinfo --sys Rescue:~ # ps -A u | grep 1359 root 1359 5.1 1.3 6048 3336 ttyS0 T 13:27 0:00 hwinfo --sys root 1361 0.0 0.2 3380 596 ttyS0 S+ 13:27 0:00 grep 1359 Rescue:~ #
Created attachment 49064 [details] vmware test program A wild guess would be the vmware check. Could you compile & run the attached simple test program? If will segfault on normal machines. Does it hang?
Created attachment 49068 [details] simpler test program This one should either segfault (normal) or at least not hang. If it hangs, it's an uml kernel problem. Independent of this, please test ~snwint/Export/hwinfo (running 'hwprobe=+sys hwinfo --sys'. It should work fine.
Both tests segfaults: Rescue:~ # ./t Segmentation fault Rescue:~ # ./t2 Segmentation fault The new hwinfo is OK: Rescue:~ # hwprobe=+sys ./hwinfo --sys 02: None 00.0: 10107 System [Created at sys.56] Unique ID: rdCR.n_7QNeEnh23 Hardware Class: system Model: "System" Config Status: cfg=new, avail=yes, need=no, active=unknown
Thanks. Could you please give the brand new hwinfo (11.23) in ~snwint/Export/hwinfo a try?
11.23 is OK: 02: None 00.0: 10107 System [Created at sys.56] Unique ID: rdCR.n_7QNeEnh23 Hardware Class: system Model: "System" Config Status: cfg=new, avail=yes, need=no, active=unknown
Thanks. Will use this fix then.