* [PATCH] sym.c: fix module_kallsyms_on_each_symbol not exported
@ 2024-06-18 16:52 tonyj
2024-06-20 1:37 ` William Cohen
0 siblings, 1 reply; 3+ messages in thread
From: tonyj @ 2024-06-18 16:52 UTC (permalink / raw)
To: systemtap; +Cc: mcermak, wcohen
module_kallsyms_on_each_symbol has never been exported (as noted by
33fae2d0107f ("This one seems simply like a non-export").
If kallsyms_on_each_symbol is exported (a revert of upstream in our
SLES case) then '!defined(STAPCONF_KALLSYMS_ON_EACH_SYMBOL_EXPORTED)'
is false and the local definition of module_kallsyms_on_each_symbol is
skipped.
This results in:
ERROR: modpost: "module_kallsyms_on_each_symbol" undefined!
Fixes: 33fae2d0107fb6166b4eac3fdffd277829849ab0
Signed-off-by: Tony Jones <tonyj@suse.de>
diff --git a/runtime/sym.c b/runtime/sym.c
index 23dd3be30..102257965 100644
--- a/runtime/sym.c
+++ b/runtime/sym.c
@@ -1180,7 +1180,8 @@ unsigned long kallsyms_lookup_name (const char *name)
}
#endif
-#if defined(STAPCONF_KALLSYMS_ON_EACH_SYMBOL) && !defined(STAPCONF_KALLSYMS_ON_EACH_SYMBOL_EXPORTED)
+#if defined(STAPCONF_KALLSYMS_ON_EACH_SYMBOL)
+#if !defined(STAPCONF_KALLSYMS_ON_EACH_SYMBOL_EXPORTED)
#ifndef KALLSYMS_H_INCLUDED
#include <linux/kallsyms.h>
#endif
@@ -1209,11 +1210,12 @@ int kallsyms_on_each_symbol(int (*fn)(void *, const char *, struct module *,
_stp_error("BUG: attempting to use unavailable kallsyms_on_each_symbol!!\n");
return 0;
}
+#endif // !defined(STAPCONF_KALLSYMS_ON_EACH_SYMBOL_EXPORTED)
+// XXX module_kallsyms_on_each_symbol has never been exported
#if LINUX_VERSION_CODE >= KERNEL_VERSION(5,12,0)
typedef typeof(&module_kallsyms_on_each_symbol) module_kallsyms_on_each_symbol_fn;
-// XXX Will be linked in place of the kernel's module_kallsyms_on_each_symbol:
#if defined(STAPCONF_KALLSYMS_6_4)
int module_kallsyms_on_each_symbol(const char *modname,
int (*fn)(void *, const char *,
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] sym.c: fix module_kallsyms_on_each_symbol not exported
2024-06-18 16:52 [PATCH] sym.c: fix module_kallsyms_on_each_symbol not exported tonyj
@ 2024-06-20 1:37 ` William Cohen
2024-06-20 18:32 ` tonyj
0 siblings, 1 reply; 3+ messages in thread
From: William Cohen @ 2024-06-20 1:37 UTC (permalink / raw)
To: tonyj, systemtap; +Cc: wcohen, mcermak
Hi,
I plan on taking a look at this patch Thursday. Do you have a pointer to the environment (distribution, specific kernel) where systemtap failed WITHOUT this patch? Thanks,
-Will
On 6/18/24 12:52, tonyj@suse.de wrote:
> module_kallsyms_on_each_symbol has never been exported (as noted by
> 33fae2d0107f ("This one seems simply like a non-export").
>
> If kallsyms_on_each_symbol is exported (a revert of upstream in our
> SLES case) then '!defined(STAPCONF_KALLSYMS_ON_EACH_SYMBOL_EXPORTED)'
> is false and the local definition of module_kallsyms_on_each_symbol is
> skipped.
>
> This results in:
> ERROR: modpost: "module_kallsyms_on_each_symbol" undefined!
>
> Fixes: 33fae2d0107fb6166b4eac3fdffd277829849ab0
> Signed-off-by: Tony Jones <tonyj@suse.de>
>
> diff --git a/runtime/sym.c b/runtime/sym.c
> index 23dd3be30..102257965 100644
> --- a/runtime/sym.c
> +++ b/runtime/sym.c
> @@ -1180,7 +1180,8 @@ unsigned long kallsyms_lookup_name (const char *name)
> }
> #endif
>
> -#if defined(STAPCONF_KALLSYMS_ON_EACH_SYMBOL) && !defined(STAPCONF_KALLSYMS_ON_EACH_SYMBOL_EXPORTED)
> +#if defined(STAPCONF_KALLSYMS_ON_EACH_SYMBOL)
> +#if !defined(STAPCONF_KALLSYMS_ON_EACH_SYMBOL_EXPORTED)
> #ifndef KALLSYMS_H_INCLUDED
> #include <linux/kallsyms.h>
> #endif
> @@ -1209,11 +1210,12 @@ int kallsyms_on_each_symbol(int (*fn)(void *, const char *, struct module *,
> _stp_error("BUG: attempting to use unavailable kallsyms_on_each_symbol!!\n");
> return 0;
> }
> +#endif // !defined(STAPCONF_KALLSYMS_ON_EACH_SYMBOL_EXPORTED)
>
> +// XXX module_kallsyms_on_each_symbol has never been exported
> #if LINUX_VERSION_CODE >= KERNEL_VERSION(5,12,0)
> typedef typeof(&module_kallsyms_on_each_symbol) module_kallsyms_on_each_symbol_fn;
>
> -// XXX Will be linked in place of the kernel's module_kallsyms_on_each_symbol:
> #if defined(STAPCONF_KALLSYMS_6_4)
> int module_kallsyms_on_each_symbol(const char *modname,
> int (*fn)(void *, const char *,
>
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] sym.c: fix module_kallsyms_on_each_symbol not exported
2024-06-20 1:37 ` William Cohen
@ 2024-06-20 18:32 ` tonyj
0 siblings, 0 replies; 3+ messages in thread
From: tonyj @ 2024-06-20 18:32 UTC (permalink / raw)
To: William Cohen; +Cc: systemtap, mcermak
On Wed, Jun 19, 2024 at 09:37:58PM -0400, William Cohen wrote:
> Hi,
> I plan on taking a look at this patch Thursday. Do you have a pointer to the environment (distribution, specific kernel) where systemtap failed WITHOUT this patch? Thanks,
Hi.
In SLE15SP5/SP6 we historically have reverted upstream 0bd476e6c671
(see: https://github.com/SUSE/kernel-source/commit/d62679f1948
It only started showing up in SP6 because we updated to stap 5.0 and
so picked up stap commit 33fae2d0107f (tags/release-5.0a~94).
I suspect the issue would show on any upstream kernel prior to
0bd476e6c671 (tags/v5.7-rc1~58^2~15)
The issue is fairly simple. module_kallsyms_on_each_symbol has never
been exported so it can't be guarded by STAPCONF_KALLSYMS_ON_EACH_SYMBOL_EXPORTED
Thanks
Tony
>
> -Will
> On 6/18/24 12:52, tonyj@suse.de wrote:
> > module_kallsyms_on_each_symbol has never been exported (as noted by
> > 33fae2d0107f ("This one seems simply like a non-export").
> >
> > If kallsyms_on_each_symbol is exported (a revert of upstream in our
> > SLES case) then '!defined(STAPCONF_KALLSYMS_ON_EACH_SYMBOL_EXPORTED)'
> > is false and the local definition of module_kallsyms_on_each_symbol is
> > skipped.
> >
> > This results in:
> > ERROR: modpost: "module_kallsyms_on_each_symbol" undefined!
> >
> > Fixes: 33fae2d0107fb6166b4eac3fdffd277829849ab0
> > Signed-off-by: Tony Jones <tonyj@suse.de>
> >
> > diff --git a/runtime/sym.c b/runtime/sym.c
> > index 23dd3be30..102257965 100644
> > --- a/runtime/sym.c
> > +++ b/runtime/sym.c
> > @@ -1180,7 +1180,8 @@ unsigned long kallsyms_lookup_name (const char *name)
> > }
> > #endif
> >
> > -#if defined(STAPCONF_KALLSYMS_ON_EACH_SYMBOL) && !defined(STAPCONF_KALLSYMS_ON_EACH_SYMBOL_EXPORTED)
> > +#if defined(STAPCONF_KALLSYMS_ON_EACH_SYMBOL)
> > +#if !defined(STAPCONF_KALLSYMS_ON_EACH_SYMBOL_EXPORTED)
> > #ifndef KALLSYMS_H_INCLUDED
> > #include <linux/kallsyms.h>
> > #endif
> > @@ -1209,11 +1210,12 @@ int kallsyms_on_each_symbol(int (*fn)(void *, const char *, struct module *,
> > _stp_error("BUG: attempting to use unavailable kallsyms_on_each_symbol!!\n");
> > return 0;
> > }
> > +#endif // !defined(STAPCONF_KALLSYMS_ON_EACH_SYMBOL_EXPORTED)
> >
> > +// XXX module_kallsyms_on_each_symbol has never been exported
> > #if LINUX_VERSION_CODE >= KERNEL_VERSION(5,12,0)
> > typedef typeof(&module_kallsyms_on_each_symbol) module_kallsyms_on_each_symbol_fn;
> >
> > -// XXX Will be linked in place of the kernel's module_kallsyms_on_each_symbol:
> > #if defined(STAPCONF_KALLSYMS_6_4)
> > int module_kallsyms_on_each_symbol(const char *modname,
> > int (*fn)(void *, const char *,
> >
>
--
Tony Jones
SUSE Kernel Performance Team
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2024-06-20 18:32 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2024-06-18 16:52 [PATCH] sym.c: fix module_kallsyms_on_each_symbol not exported tonyj
2024-06-20 1:37 ` William Cohen
2024-06-20 18:32 ` tonyj
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).