After building 2.6.10-rc1, i was unable to load my "lufs" module due to
an unknown symbol error (kill_proc_info). When I examined the
2.6.10-rc1 patch, I noticed that "EXPORT_SYMBOL(kill_proc_info);" was
removed from signal.c. With the following patch, I was able to resolve
my problem, but I am not sure if this is the correct method. Is there a
reason why the kill_proc_info symbol is no longer exported?
--- linux/kernel/signal.c.original 2004-10-23 02:12:32.000000000 -0700
+++ linux/kernel/signal.c 2004-10-23 01:37:36.000000000 -0700
@@ -1939,6 +1939,7 @@
EXPORT_SYMBOL(force_sig);
EXPORT_SYMBOL(kill_pg);
EXPORT_SYMBOL(kill_proc);
+EXPORT_SYMBOL(kill_proc_info);
EXPORT_SYMBOL(ptrace_notify);
EXPORT_SYMBOL(send_sig);
EXPORT_SYMBOL(send_sig_info);
On Sat, Oct 23, 2004 at 02:21:22AM -0700, C.Y.M wrote:
> After building 2.6.10-rc1, i was unable to load my "lufs" module due to
> an unknown symbol error (kill_proc_info). When I examined the
> 2.6.10-rc1 patch, I noticed that "EXPORT_SYMBOL(kill_proc_info);" was
> removed from signal.c. With the following patch, I was able to resolve
> my problem, but I am not sure if this is the correct method. Is there a
> reason why the kill_proc_info symbol is no longer exported?
Because it's not an API you should be using.
On Sat, 2004-10-23 at 11:21, C.Y.M wrote:
> After building 2.6.10-rc1, i was unable to load my "lufs" module due to
> an unknown symbol error (kill_proc_info). When I examined the
the big question is why lufs uses kill_proc_info in the first place...
Christoph Hellwig wrote:
> On Sat, Oct 23, 2004 at 02:21:22AM -0700, C.Y.M wrote:
>
>>After building 2.6.10-rc1, i was unable to load my "lufs" module due to
>>an unknown symbol error (kill_proc_info). When I examined the
>>2.6.10-rc1 patch, I noticed that "EXPORT_SYMBOL(kill_proc_info);" was
>>removed from signal.c. With the following patch, I was able to resolve
>>my problem, but I am not sure if this is the correct method. Is there a
>>reason why the kill_proc_info symbol is no longer exported?
>
>
> Because it's not an API you should be using.
>
>
Is there an alternative?
On Sat, Oct 23, 2004 at 03:04:47AM -0700, C.Y.M wrote:
> Christoph Hellwig wrote:
> >On Sat, Oct 23, 2004 at 02:21:22AM -0700, C.Y.M wrote:
> >
> >>After building 2.6.10-rc1, i was unable to load my "lufs" module due to
> >>an unknown symbol error (kill_proc_info). When I examined the
> >>2.6.10-rc1 patch, I noticed that "EXPORT_SYMBOL(kill_proc_info);" was
> >>removed from signal.c. With the following patch, I was able to resolve
> >>my problem, but I am not sure if this is the correct method. Is there a
> >>reason why the kill_proc_info symbol is no longer exported?
> >
> >
> >Because it's not an API you should be using.
> >
> >
> Is there an alternative?
Maybe you could explain what you're actually trying to do at a higher
level first.
Christoph Hellwig wrote:
> On Sat, Oct 23, 2004 at 03:04:47AM -0700, C.Y.M wrote:
>
>>Christoph Hellwig wrote:
>>
>>>On Sat, Oct 23, 2004 at 02:21:22AM -0700, C.Y.M wrote:
>>>
>>>
>>>>After building 2.6.10-rc1, i was unable to load my "lufs" module due to
>>>>an unknown symbol error (kill_proc_info). When I examined the
>>>>2.6.10-rc1 patch, I noticed that "EXPORT_SYMBOL(kill_proc_info);" was
>>>>removed from signal.c. With the following patch, I was able to resolve
>>>>my problem, but I am not sure if this is the correct method. Is there a
>>>>reason why the kill_proc_info symbol is no longer exported?
>>>
>>>
>>>Because it's not an API you should be using.
>>>
>>>
>>
>>Is there an alternative?
>
>
> Maybe you could explain what you're actually trying to do at a higher
> level first.
>
I have been using the lufs module in combination with autofs to be able
to automount ftp sites on the fly. But, if lufs is broken due to the
lack of the "kill_proc_info" symbol being available, perhaps I will just
remove lufs and find another way (or wait until there is some kind of
patch for lufs).
On Sat, Oct 23, 2004 at 04:17:42AM -0700, C.Y.M wrote:
> >>Is there an alternative?
> >
> >
> >Maybe you could explain what you're actually trying to do at a higher
> >level first.
> >
> I have been using the lufs module in combination with autofs to be able
> to automount ftp sites on the fly. But, if lufs is broken due to the
> lack of the "kill_proc_info" symbol being available, perhaps I will just
> remove lufs and find another way (or wait until there is some kind of
> patch for lufs).
Well, that's _too_ highlevel :) The question is what lufs is trying to
archive by using the module.