* [PATCH] module_name()
@ 2002-11-12 17:32 Rusty Russell
2002-11-12 20:39 ` Robert Love
2002-11-12 22:03 ` David S. Miller
0 siblings, 2 replies; 6+ messages in thread
From: Rusty Russell @ 2002-11-12 17:32 UTC (permalink / raw)
To: davem; +Cc: linux-kernel
I prefer this: it also has the advantage of ensuring the name of
built-in modules is consistent across the kernel.
Thanks for the report,
Rusty.
--
Anyone who quotes me in their sig is an idiot. -- Rusty Russell.
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal working-2.5.47-module-alias/crypto/api.c working-2.5.47-modname/crypto/api.c
--- working-2.5.47-module-alias/crypto/api.c 2002-11-11 20:00:55.000000000 +1100
+++ working-2.5.47-modname/crypto/api.c 2002-11-13 04:30:48.000000000 +1100
@@ -263,8 +263,7 @@ static int c_show(struct seq_file *m, vo
struct crypto_alg *alg = (struct crypto_alg *)p;
seq_printf(m, "name : %s\n", alg->cra_name);
- seq_printf(m, "module : %s\n", alg->cra_module ?
- alg->cra_module->name : "[static]");
+ seq_printf(m, "module : %s\n", module_name(alg->cra_module));
seq_printf(m, "blocksize : %u\n", alg->cra_blocksize);
switch (alg->cra_flags & CRYPTO_ALG_TYPE_MASK) {
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal working-2.5.47-module-alias/include/linux/module.h working-2.5.47-modname/include/linux/module.h
--- working-2.5.47-module-alias/include/linux/module.h 2002-11-12 22:43:55.000000000 +1100
+++ working-2.5.47-modname/include/linux/module.h 2002-11-13 04:31:16.000000000 +1100
@@ -293,6 +293,13 @@ static inline void module_put(struct mod
#endif /* CONFIG_MODULE_UNLOAD */
+static inline char *module_name(struct module *module)
+{
+ if (module)
+ return module->name;
+ return "[built-in]";
+}
+
#define __unsafe(mod) \
do { \
if (mod && !(mod)->unsafe) { \
@@ -315,6 +322,10 @@ do { \
#define try_module_get(module) 1
#define module_put(module) do { } while(0)
+static inline char *module_name(struct module *module)
+{
+ Return "[built-in]";
+}
#define __unsafe(mod)
#endif /* CONFIG_MODULES */
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] module_name()
2002-11-12 17:32 [PATCH] module_name() Rusty Russell
@ 2002-11-12 20:39 ` Robert Love
2002-11-12 22:00 ` David S. Miller
2002-11-12 22:03 ` David S. Miller
1 sibling, 1 reply; 6+ messages in thread
From: Robert Love @ 2002-11-12 20:39 UTC (permalink / raw)
To: Rusty Russell; +Cc: davem, linux-kernel
On Tue, 2002-11-12 at 12:32, Rusty Russell wrote:
> +static inline char *module_name(struct module *module)
> +{
> + Return "[built-in]";
> +}
s/Return/return/ ? ;-)
Robert Love
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] module_name()
2002-11-12 20:39 ` Robert Love
@ 2002-11-12 22:00 ` David S. Miller
0 siblings, 0 replies; 6+ messages in thread
From: David S. Miller @ 2002-11-12 22:00 UTC (permalink / raw)
To: rml; +Cc: rusty, linux-kernel
From: Robert Love <rml@tech9.net>
Date: 12 Nov 2002 15:39:14 -0500
On Tue, 2002-11-12 at 12:32, Rusty Russell wrote:
> +static inline char *module_name(struct module *module)
> +{
> + Return "[built-in]";
> +}
s/Return/return/ ? ;-)
I'd rather get a NULL for built-in, honestly.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] module_name()
2002-11-12 17:32 [PATCH] module_name() Rusty Russell
2002-11-12 20:39 ` Robert Love
@ 2002-11-12 22:03 ` David S. Miller
2002-11-13 10:04 ` Rusty Russell
1 sibling, 1 reply; 6+ messages in thread
From: David S. Miller @ 2002-11-12 22:03 UTC (permalink / raw)
To: rusty; +Cc: linux-kernel
From: Rusty Russell <rusty@rustcorp.com.au>
Date: Wed, 13 Nov 2002 04:32:58 +1100
I prefer this: it also has the advantage of ensuring the name of
built-in modules is consistent across the kernel.
I'd rather get NULL for built-in, this also eliminates
the issue of having umpteen "[built-in]" string copies in
the kernel since this is expanded by an inline.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] module_name()
2002-11-12 22:03 ` David S. Miller
@ 2002-11-13 10:04 ` Rusty Russell
2002-11-13 10:56 ` David S. Miller
0 siblings, 1 reply; 6+ messages in thread
From: Rusty Russell @ 2002-11-13 10:04 UTC (permalink / raw)
To: David S. Miller; +Cc: linux-kernel
In message <20021112.140357.81690208.davem@redhat.com> you write:
> From: Rusty Russell <rusty@rustcorp.com.au>
> Date: Wed, 13 Nov 2002 04:32:58 +1100
>
> I prefer this: it also has the advantage of ensuring the name of
> built-in modules is consistent across the kernel.
>
> I'd rather get NULL for built-in
I'd rather not. If you want that check, check for module == NULL.
This makes it trivially consistent across proc output, which is the
main use of module name outside module.c.
> this also eliminates the issue of having umpteen "[built-in]" string
> copies in the kernel since this is expanded by an inline.
Ick, yes. Turned into macros, and changed to "kernel" since
exec_domain.c already uses that.
Thoughts?
Rusty.
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal linux-2.5-bk/crypto/api.c working-2.5-bk-modname/crypto/api.c
--- linux-2.5-bk/crypto/api.c 2002-11-11 20:00:55.000000000 +1100
+++ working-2.5-bk-modname/crypto/api.c 2002-11-13 20:49:52.000000000 +1100
@@ -263,8 +263,7 @@ static int c_show(struct seq_file *m, vo
struct crypto_alg *alg = (struct crypto_alg *)p;
seq_printf(m, "name : %s\n", alg->cra_name);
- seq_printf(m, "module : %s\n", alg->cra_module ?
- alg->cra_module->name : "[static]");
+ seq_printf(m, "module : %s\n", module_name(alg->cra_module));
seq_printf(m, "blocksize : %u\n", alg->cra_blocksize);
switch (alg->cra_flags & CRYPTO_ALG_TYPE_MASK) {
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal linux-2.5-bk/include/linux/module.h working-2.5-bk-modname/include/linux/module.h
--- linux-2.5-bk/include/linux/module.h 2002-11-13 18:54:55.000000000 +1100
+++ working-2.5-bk-modname/include/linux/module.h 2002-11-13 20:56:51.000000000 +1100
@@ -242,6 +242,13 @@ static inline void module_put(struct mod
#endif /* CONFIG_MODULE_UNLOAD */
+/* This is a #define so the string doesn't get put in every .o file */
+#define module_name(mod) \
+({ \
+ struct module *__mod = (mod); \
+ __mod ? __mod->name : "kernel"; \
+})
+
#define __unsafe(mod) \
do { \
if (mod && !(mod)->unsafe) { \
@@ -265,6 +272,8 @@ do { \
#define try_module_get(module) 1
#define module_put(module) do { } while(0)
+#define module_name(mod) "kernel"
+
#define __unsafe(mod)
#endif /* CONFIG_MODULES */
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal linux-2.5-bk/kernel/exec_domain.c working-2.5-bk-modname/kernel/exec_domain.c
--- linux-2.5-bk/kernel/exec_domain.c 2002-11-13 18:54:56.000000000 +1100
+++ working-2.5-bk-modname/kernel/exec_domain.c 2002-11-13 20:57:19.000000000 +1100
@@ -210,13 +210,8 @@ get_exec_domain_list(char *page)
read_lock(&exec_domains_lock);
for (ep = exec_domains; ep && len < PAGE_SIZE - 80; ep = ep->next)
len += sprintf(page + len, "%d-%d\t%-16s\t[%s]\n",
- ep->pers_low, ep->pers_high, ep->name,
-#ifdef CONFIG_MODULES
- ep->module ? ep->module->name : "kernel"
-#else
- "kernel"
-#endif
- );
+ ep->pers_low, ep->pers_high, ep->name,
+ module_name(ep->module));
read_unlock(&exec_domains_lock);
return (len);
}
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal linux-2.5-bk/net/ipv4/netfilter/ip_nat_helper.c working-2.5-bk-modname/net/ipv4/netfilter/ip_nat_helper.c
--- linux-2.5-bk/net/ipv4/netfilter/ip_nat_helper.c 2002-11-13 18:54:56.000000000 +1100
+++ working-2.5-bk-modname/net/ipv4/netfilter/ip_nat_helper.c 2002-11-13 20:59:30.000000000 +1100
@@ -361,8 +361,6 @@ helper_cmp(const struct ip_nat_helper *h
return ip_ct_tuple_mask_cmp(tuple, &helper->tuple, &helper->mask);
}
-#define MODULE_MAX_NAMELEN 32
-
int ip_nat_helper_register(struct ip_nat_helper *me)
{
int ret = 0;
@@ -374,14 +372,13 @@ int ip_nat_helper_register(struct ip_nat
&& ct_helper->me) {
__MOD_INC_USE_COUNT(ct_helper->me);
} else {
-#ifdef CONFIG_MODULES
/* We are a NAT helper for protocol X. If we need
* respective conntrack helper for protoccol X, compute
* conntrack helper name and try to load module */
- char name[MODULE_MAX_NAMELEN];
- const char *tmp = me->me->name;
+ char name[MODULE_NAME_LEN];
+ const char *tmp = module_name(me->me);
- if (strlen(tmp) + 6 > MODULE_MAX_NAMELEN) {
+ if (strlen(tmp) + 6 > MODULE_NAME_LEN) {
printk("%s: unable to "
"compute conntrack helper name "
"from %s\n", __FUNCTION__, tmp);
@@ -404,7 +401,6 @@ int ip_nat_helper_register(struct ip_nat
"module loader support\n", name);
return -EBUSY;
#endif
-#endif
}
}
WRITE_LOCK(&ip_nat_lock);
--
Anyone who quotes me in their sig is an idiot. -- Rusty Russell.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] module_name()
2002-11-13 10:04 ` Rusty Russell
@ 2002-11-13 10:56 ` David S. Miller
0 siblings, 0 replies; 6+ messages in thread
From: David S. Miller @ 2002-11-13 10:56 UTC (permalink / raw)
To: rusty; +Cc: linux-kernel
From: Rusty Russell <rusty@rustcorp.com.au>
Date: Wed, 13 Nov 2002 21:04:10 +1100
> this also eliminates the issue of having umpteen "[built-in]" string
> copies in the kernel since this is expanded by an inline.
Ick, yes. Turned into macros, and changed to "kernel" since
exec_domain.c already uses that.
Thoughts?
Ok, since when mod is NULL crypto.h won't even invoke module_name()
anyways :-)
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2002-11-13 10:51 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2002-11-12 17:32 [PATCH] module_name() Rusty Russell
2002-11-12 20:39 ` Robert Love
2002-11-12 22:00 ` David S. Miller
2002-11-12 22:03 ` David S. Miller
2002-11-13 10:04 ` Rusty Russell
2002-11-13 10:56 ` David S. Miller
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 NNTP newsgroup(s).