On martedì 27 marzo 2007, Pravin wrote: > > Hmm, I published a 2.6.18 -bs patchset time ago with some bugfixes, and > > some other were merged into -stable, so you may want to try these. CKRM > > will still work with them (98 % sure).
> I tried both of the following patches on vanilla 2.6.18 kernel. > (Without CKRM patch) > http://www.user-mode-linux.org/~blaisorblade/patches/guest/uml-2.6.18.1-bb2 >/uml-2.6.18.1-bb2.patch.bz2 and > http://www.user-mode-linux.org/~blaisorblade/patches/guest/uml-2.6.18-bb1/u >ml-2.6.18-bb1.patch.bz2 > But still it UML didnt worked. > I used GDB to locate the point where it is failing, so i got following. > {{{ > (gdb) bt > #0 __const_udelay (usecs=4295000) at arch/um/sys-i386/delay.c:36 > #1 0x0806c878 in panic (fmt=0x804e79c "\203�\024h;�\035\b��\226") at > kernel/panic.c:137 > #2 0x0804e79c in check_ptrace () > #3 0x0806384b in os_early_checks () at include/linux/timer.h:41 > #4 0x0805c1c3 in linux_main (argc=1, argv=0xafdc3e24) at > arch/um/kernel/um_arch.c:352 > #5 0x08061fb2 in main () at include/linux/timer.h:41 > (gdb) > }}} > > So, what I concluded from this that, panic is getting called from > "check_ptrace" function. > > To find out why exactly panic is getting called, I added few printf > statements in "check_ptrace" function in > "arch/um/os-Linux/start_up.c". > I managed to find out that panic is getting called from > {{{ > if(!WIFSTOPPED(status) || (WSTOPSIG(status) != (SIGTRAP|0x80))) > panic("check_ptrace : expected (SIGTRAP|0x80), " > "got status = %d", status); > }}} > It was not calling panic in first iteration but it was calling in > second iteration inside while loop, with panic statement as > {{{ > check_ptrace : expected (SIGTRAP|0x80), got status = 256 > }}} > At first iteration, the value of status was "34175". My program decodes it as SIGTRAP|0x80, which is correct. $ ./waitstatus 34175 WSTOPSIG(status) is 0x80|5 The signal is: Trace/breakpoint trap > I suppose that it should have passed the condition in following "if > statement" and should have taken a break in first iteration only. Yes. > /////////////////////////////////////////////////////////////////////////// >// syscall = ptrace(PTRACE_PEEKUSR, pid, PT_SYSCALL_NR_OFFSET,0); > if(syscall == __NR_getpid){ > n = ptrace(PTRACE_POKEUSR, pid, PT_SYSCALL_NR_OFFSET, > __NR_getppid); > if(n < 0) > panic("check_ptrace : failed to modify system " > "call, errno = %d", errno); > break ; > } > /////////////////////////////////////////////////////////////////////////// >// But the value of syscall in first iteration was 252 but it was > expected to be 20 (i.e. __NR_getpid ). > Value 252 represented "__NR_exit_group" Hmm, there are some glibc which cache the result of getpid(), so we use syscall(getpid). It seems this is happening again here, but it's strange. I looked in 2.6.18 source code and the bug had already been fixed. Can you test if 2.6.17 compiles for you and has this bug? Now, again, I need to know: - distribution - host kernel - glibc version used - whether the host processor/distro is a 64-bit one The only difference in source code between 2.6.18 and 2.6.19 which may be related (but I have serious doubts, there shouldn't be different definitions) is this: diff --git a/arch/um/os-Linux/process.c b/arch/um/os-Linux/process.c index b98d3ca..c692a19 100644 --- a/arch/um/os-Linux/process.c +++ b/arch/um/os-Linux/process.c @@ -7,10 +7,10 @@ #include <stdio.h> #include <errno.h> #include <signal.h> -#include <linux/unistd.h> #include <sys/mman.h> #include <sys/wait.h> #include <sys/mman.h> +#include <sys/syscall.h> #include "ptrace_user.h" #include "os.h" #include "user.h" > I am not able to proceed further from this point in debugging this problem. That's already a lot > > nothing of interest. Have you double-checked your compile-time options? > > > > > > Looked at what's happening with strace? > > on using strace on uml i got following. But I am not able to make any > sense out of it :-( Argh, strace -f breaks UML, but would be what is needed, so strace is useless. Btw, adding stderr=1 should make UML print its error message in such cases. -- Inform me of my mistakes, so I can add them to my list! Paolo Giarrusso, aka Blaisorblade http://www.user-mode-linux.org/~blaisorblade ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ User-mode-linux-user mailing list User-mode-linux-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/user-mode-linux-user
