Skip to contentSkip to conversation

coredump - as root not only if euid switched

PATCH Peter Waechtler 2 participants 2 messages Apr 22, 2004

lore
Share & tools
lore
#0 lore raw
Quoted 22 lines · Andrew Morton
Quoted 7 lines · Peter W�chtler

While it's more secure to not dump core at all if the program has
switched euid, it's also very unpractical. Since only programs started from root, being setuid root or have CAP_SETUID it's far more practical to dump as root.root mode 600. This is the bahavior of Solaris.

Quoted 7 lines

The current implementation does not ensure that an existing core file is only readable as root, i.e. after dumping the ownership and mode is unchanged.

Besides mm->dumpable to avoid recursive core dumps, on setuid files the dumpable flag still prevents a core dump while seteuid & co will result in a core only readable as root.

It's a bit sad to add another function call level to sys_unlink() simply because the core dumping code needs it.

Is it not possible to call sys_unlink() directly from there? Something like

long kernel_unlink(const char *name)

Quoted 9 lines
{
	mm_segment_t old_fs = get_fs();
	long ret;

	set_fs(KERNEL_DS);
	ret = sys_unlink(name);
	set_fs(old_fs);
	return ret;
}	

And you're asking me? ;)
While getname() has a check for user/kernelspace - do you really care about "the overhead" for a function call level with 1 argument? Uhm, probably if you even care to write this..

What is the cost for switching the segments?
But I agree that sys_unlink should be the fast call and dumping core is the exception :)

would fastcall do_unlink() help? I guess the arg is then passed in a register

#1 lore raw
Quoted 8 lines · Peter Waechtler
Quoted 8 lines

Is it not possible to call sys_unlink() directly from there? Something like

long kernel_unlink(const char *name)

{
	mm_segment_t old_fs = get_fs();
	long ret;
Quoted 7 lines
Quoted 5 lines
	set_fs(KERNEL_DS);
	ret = sys_unlink(name);
	set_fs(old_fs);
	return ret;
}	

And you're asking me? ;)

Well someone has to know ;)

Quoted 3 lines

While getname() has a check for user/kernelspace - do you really care about "the overhead" for a function call level with 1 argument? Uhm, probably if you even care to write this..

It's very small, but heck, we unlink a lot more often than we dump core. Most of us, that is.

Quoted 1 line

What is the cost for switching the segments?

Teeny.

Quoted 5 lines

But I agree that sys_unlink should be the fast call and dumping core is the exception :)

would fastcall do_unlink() help? I guess the arg is then passed in a register

I've never been able to measure any size or space benefit for fastcall, and we do it via compiler options kernel-wide nowadays.

The above will work fine. You can probably just open-code it at the place where you're unlinking the file.

(why are you trying to unlink the old file anyway?)