Re: hook SIGSEGV
On 2014-05-10 21:46:01 (+0800), net.study@gmail.com net.study@gmail.com wrote: I want to know is it possible to hook SIGSEGV to restart the thread which the signal is sent to,without restart the whole process? And record the place where has caused this signal? Yes, as others have already pointed out, you can hook SIGSEGV like any other signal. You're not going to be able to save the process any more, but you can still collect some useful information. I've found it very useful to have a SIGSEGV (and SIGPIPE, SIGABRT, SIGFPE, SIGILL) handler which logs a backtrace (look at 'man backtrace') to syslog. Very useful for debugging on targets where core dumps are impractical. Others have also pointed out that it might no longer be safe to call printf() or malloc() there. That's true, but usually it's OK, and if it turns out that it wasn't ... Well, you were crashing anyway. Regards, Kristof ___ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
a question on nocache pool address which seems to have ended up as a nice comment?
Hi, I was analyzing arch/sparc/mm/srmmu.c (linux 3.3) In our system, the pgd table sits at c800 and the entry for c8000 is at c8000320. (upper 8 bits is the index and *4 is the index address) previously, I got srmmu_nocache_pool = c054. Does anybody know why __nocache_fix(0xc8320) is c0540320? This is equivalent to saying that nocache region 0xc800 is mapped to 0xc054 which is physical address 0x6054. I'm confused here.. I guess the virtual address 0xc800 which is in nocache region is rebased to previously allocated nocache physcial address 0x6054 which is then equivalent to virtual address 0xc054. Having said that, it sounds perfectly correct. another example of realizing something while asking.. Correct me if I'm wrong. regards, Chan ___ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
Deleting a line from a file
Hi, I have a file that has entries for different absolute path on separate lines. eg: /opt/new1 /opt/new2 I need to delete an entry from this file for a given path, for which I am using sed. sed -i 's#^/opt/new1.*$##g' file_entries.txt However this is leaving blank line in between, which I don't want. Can someone help me to do this without leaving blank lines in between. Regards, Saket Sinha ___ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
Re: Deleting a line from a file
Hi! The original mail is off-topic as it has nothing to do wotj the Linux kernel development as such, but: On Mit, 2014-05-14 at 19:58 +0530, Saket Sinha wrote: [...] I have a file that has entries for different absolute path on separate lines. eg: /opt/new1 /opt/new2 I need to delete an entry from this file for a given path, for which I am using sed. sed -i 's#^/opt/new1.*$##g' file_entries.txt However this is leaving blank line in between, which I don't want. Can someone help me to do this without leaving blank lines in between. sed -i 's#^/opt/new1.*$#d' file_entries.txt should do it Bernd -- Bernd Petrovitsch Email : be...@petrovitsch.priv.at LUGA : http://www.luga.at ___ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
Re: Deleting a line from a file
On Wed, 14 May 2014 16:34:20 +0200, Bernd Petrovitsch said: sed -i 's#^/opt/new1.*$#d' file_entries.txt You don't even need the leading 's'. Just /pattern/d is sufficient. (And you can even do stuff like /pat1/,/pat2/s/old/new/ which will change 'old' to 'new', but only from a line that contains pat1 up to a line that contains pat2, and *not* changing it before pat1 or after pat2) pgpAcbxZ3MSYc.pgp Description: PGP signature ___ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
Re: Deleting a line from a file
Please find my response inline-
On Wed, May 14, 2014 at 8:04 PM, Bernd Petrovitsch
be...@petrovitsch.priv.at wrote:
Hi!
The original mail is off-topic as it has nothing to do wotj the Linux
kernel development as such, but:
I wanted to isolate the problem hence did not give the full context.
Actually I have written a file-system utility which mounts my
filesystem on a specific path and update it in /etc/fstab. When I
umount it, I delete the path from the /etc/fstab.
I had earlier written a C function to do that but it is corrupting the fstab.
SO now I am trying to run the above sed command this way
system(sed -i 's#^/opt/new1.*$##g' /etc/fstab).
The C function that was currupting the filesystem is listed below -
int removeEntryFromFSTAB(const char * fullPath, const char * fileName)
{
FILE *tabFileOld = NULL;
FILE *tabFileNew = NULL;
struct mntent *m;
char newFileName[PATH_MAX];
int rc = -1;
tabFileOld = setmntent( fileName, r ); // Open for writing now
if (tabFileOld == NULL )
goto end;
tabFileNew = setmntent(newFileName, w);
if (tabFileNew == NULL )
goto end;
while ((m = getmntent(tabFileOld)) != NULL)
{
if (tcscmp(MY_FS_TYPE, m-mnt_type) == 0)
{
if ((tcscmp(fullPath, m-mnt_dir) == 0))
continue;
}
if (addmntent(tabFileNew, m) != 0)
goto end;
}
endmntent(tabFileOld);
endmntent(tabFileNew);
tabFileNew = NULL;
tabFileOld = NULL;
rename(newFileName, fileName))
rc = 0;
end:
if (tabFileNew != NULL)
endmntent(tabFileNew);
if (tabFileOld != NULL)
endmntent(tabFileOld);
sync();
return rc;
}
Kindly let me know, if you think there is a better way.
Regards,
Saket Sinha
___
Kernelnewbies mailing list
Kernelnewbies@kernelnewbies.org
http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
Re: Deleting a line from a file
On Wed, 14 May 2014 21:27:06 +0530, Saket Sinha said: char newFileName[PATH_MAX]; tabFileNew = setmntent(newFileName, w); And what is the new file name? You have random trash on the stack here. (Note that this is C 101 - if you can't debug this on your own, you probably shouldn't be messing with filesystem code until you have more C experience) pgpgfLmgh_JZL.pgp Description: PGP signature ___ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
Re: Deleting a line from a file
Please find response inline. On Wed, May 14, 2014 at 9:44 PM, valdis.kletni...@vt.edu wrote: On Wed, 14 May 2014 21:27:06 +0530, Saket Sinha said: char newFileName[PATH_MAX]; tabFileNew = setmntent(newFileName, w); And what is the new file name? You have random trash on the stack here. int removeEntryFromFSTAB(const char * fullPath, const char * fileName); I am sending /etc/fstab in fileName to this function and the path to be deleted in fullPath NOW char newFileName[PATH_MAX]; .. ... rename(newFileName, fileName)); I am taking a new file, writing my entries to it and then replacing it with original file. I don't find anything wrong with that (Note that this is C 101 - if you can't debug this on your own, you probably shouldn't be messing with filesystem code until you have more C experience) Sorry but I haven't been able to debug that. I admit it. Regards, Saket Sinha ___ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
Re: Deleting a line from a file
On Wed, 14 May 2014 22:13:51 +0530, Saket Sinha said:
I am sending /etc/fstab in fileName to this function and the path to
be deleted in fullPath
OK.
char newFileName[PATH_MAX];
This lives on your function call stack. As such, it contains whatever
was in that memory until you change it. And you never change it.
tabFileOld = setmntent( fileName, r ); // Open for writing now
The comment indicates there's some confusion going on. Not a good sign
tabFileNew = setmntent(newFileName, w);
if (tabFileNew == NULL )
goto end;
This probably doesn't do what you think it does. What value of 'newFileName'
is used to create the new file? For bonus points, what guarantees that
it points to an object that's on the same filesystem as your old filename?
(This will actually fail to throw an error as long as newFileName[0] isn't
a '\0'. But that's different from actually working properly)
And you *really* want to get this issue well-understood before hacking kernel
code, as the kernel doesn't do very much error checking. You make this sort of
error in kernel code, you will crash the machine - if you're lucky. If you're
unlucky, the system will silently corrupt entire filesystems on you to the
point where fsck won't help. (Want *real* debugging fun? Get your system
to a state where /boot fsck's just fine, but something has silently overwritten
several data blocks in /boot/vmlinuz or /lib/ld-linux.so. *That* will keep
you busy for a while.. :)
if (tcscmp(MY_FS_TYPE, m-mnt_type) == 0)
You want to use strncmp() or similar here. And you want to learn why
you shouldn't be using tcscmp() (Hint: are either MY_FS_TYPE or m-mnt_type
allowed to be UTF-8 strings? :)
rename(newFileName, fileName))
rc = 0;
You probaby want to check the return code:
rc = rename(newFileName, fileName);
if (!rc) { perror() or something
as you're *very* likely to get EXDEV as an error due to previous code.
You're also missing error checking on every single endmntent() call - what
happens if one of those fails (which *can* happen)?
sync();
You don't want to do that. You should have done an fsync() (except that's
technically wrong as you're mixing it with stdio and not setting it to
unbuffered) or a syncfs() instead.
return rc;
This fails to capture a number of potential errors.
pgpvij8kG0aVH.pgp
Description: PGP signature
___
Kernelnewbies mailing list
Kernelnewbies@kernelnewbies.org
http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
Re: hook SIGSEGV
On Wed, May 14, 2014 at 4:14 PM, Kristof Provost kris...@sigsegv.be wrote: On 2014-05-10 21:46:01 (+0800), net.study@gmail.com net.study@gmail.com wrote: I want to know is it possible to hook SIGSEGV to restart the thread which the signal is sent to,without restart the whole process? And record the place where has caused this signal? Yes, as others have already pointed out, you can hook SIGSEGV like any other signal. You're not going to be able to save the process any more, but you can still collect some useful information. I've found it very useful to have a SIGSEGV (and SIGPIPE, SIGABRT, SIGFPE, SIGILL) handler which logs a backtrace (look at 'man backtrace') to syslog. Very useful for debugging on targets where core dumps are impractical. Others have also pointed out that it might no longer be safe to call printf() or malloc() there. That's true, but usually it's OK, and if it turns out that it wasn't ... Well, you were crashing anyway. Well, not anyway: you still should be able to take a longjmp out of the signal handler to a safe place. -- Thanks. -- Max ___ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
Re: confused kernel compressed code link address
hi Arn Ks, thx for your reply. So there is no global variables during the decompress process of kernel. best regards 2014-05-12 14:06 GMT+08:00 Arun KS getaru...@gmail.com: Hi Chuck Huang, On Thu, May 8, 2014 at 5:25 PM, Chuck Huang chuck.ker...@gmail.com wrote: Dear all, Now i am looking into kernel decompressed code, arch/arm/boot/compressed/head.S As i objdump arch/arm/boot/compressed/vmlinux, the start section address is 0x0. vmlinux.S 373789| Disassembly of section .text: 373790| 373791| start: 373792| #endif if the code run in 0x8000 (which is DDR base address), but the link address is zero, how it works ? The compressed/head.S code is position independent code. It can be loaded anywhere in RAM. There will not be reference to any labels directly. All the address references are done relative to PC. Thanks, Arun best regards ___ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies ___ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
packet drop
Hi All, I am sending packet between two back to back connected machine.when I use same network (192.168.7.x) packet reach upto NF_IP_LOCAL_IN (IPTABLES INPUT) hook.It is accepted. But when I spoof ip address network(192.168.8.x) it pass NF_IP_PRE_ROUTING hook but dropped before NF_IP_LOCAL_IN hook. case 1: 192.168.7.12 -- 192.168.7.19 Accepted. case 2: 192.168.8.25 -- 192.168.7.19 Dropped. I didn't get why it dropped ? How network change effect on routing inside kernel tcp/ip stack ? Regards Varun ___ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
