diff mbox

[RFC] module: allow multiple calls to MODULE_DEVICE_TABLE() per module

Message ID 1390849795-2155-1-git-send-email-teg@jklm.no (mailing list archive)
State New, archived
Headers show

Commit Message

Tom Gundersen Jan. 27, 2014, 7:09 p.m. UTC
Commit 78551277e4df5: "Input: i8042 - add PNP modaliases" had a bug, where the
second call to MODULE_DEVICE_TABLE() overrode the first resulting in not all
the modaliases being exposed.

This fixes the problem by including the name of the device_id table in the
__mod_*_device_table alias, allowing us to export several device_id tables
per module.

Suggested-by: Kay Sievers <kay@vrfy.org>
Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Rusty Russell <rusty@rustcorp.com.au>
---
 include/linux/module.h   |  2 +-
 scripts/mod/file2alias.c | 14 +++++++++-----
 2 files changed, 10 insertions(+), 6 deletions(-)

Comments

Greg Kroah-Hartman Jan. 27, 2014, 7:54 p.m. UTC | #1
On Mon, Jan 27, 2014 at 08:09:55PM +0100, Tom Gundersen wrote:
> Commit 78551277e4df5: "Input: i8042 - add PNP modaliases" had a bug, where the
> second call to MODULE_DEVICE_TABLE() overrode the first resulting in not all
> the modaliases being exposed.
> 
> This fixes the problem by including the name of the device_id table in the
> __mod_*_device_table alias, allowing us to export several device_id tables
> per module.
> 
> Suggested-by: Kay Sievers <kay@vrfy.org>
> Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> Cc: Rusty Russell <rusty@rustcorp.com.au>
> ---
>  include/linux/module.h   |  2 +-
>  scripts/mod/file2alias.c | 14 +++++++++-----
>  2 files changed, 10 insertions(+), 6 deletions(-)

Ah, very nice, I've wanted this for a while now, it would make a number
of different drivers much smaller and simpler to add new device ids to
(no multiple lists of ids, one for the module loader and one for the
sub-driver that is in the single file.)

If this doesn't break any userspace tools, I have no objection to it at
all:

Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
--
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Dmitry Torokhov Jan. 27, 2014, 8:30 p.m. UTC | #2
On Mon, Jan 27, 2014 at 11:54:22AM -0800, Greg Kroah-Hartman wrote:
> On Mon, Jan 27, 2014 at 08:09:55PM +0100, Tom Gundersen wrote:
> > Commit 78551277e4df5: "Input: i8042 - add PNP modaliases" had a bug, where the
> > second call to MODULE_DEVICE_TABLE() overrode the first resulting in not all
> > the modaliases being exposed.
> > 
> > This fixes the problem by including the name of the device_id table in the
> > __mod_*_device_table alias, allowing us to export several device_id tables
> > per module.
> > 
> > Suggested-by: Kay Sievers <kay@vrfy.org>
> > Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
> > Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> > Cc: Rusty Russell <rusty@rustcorp.com.au>
> > ---
> >  include/linux/module.h   |  2 +-
> >  scripts/mod/file2alias.c | 14 +++++++++-----
> >  2 files changed, 10 insertions(+), 6 deletions(-)
> 
> Ah, very nice, I've wanted this for a while now, it would make a number
> of different drivers much smaller and simpler to add new device ids to
> (no multiple lists of ids, one for the module loader and one for the
> sub-driver that is in the single file.)
> 
> If this doesn't break any userspace tools, I have no objection to it at
> all:
> 
> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Looks great to me as long as module tools keep working.

Thanks.
Rusty Russell Jan. 28, 2014, 1:34 a.m. UTC | #3
Greg Kroah-Hartman <gregkh@linuxfoundation.org> writes:
> On Mon, Jan 27, 2014 at 08:09:55PM +0100, Tom Gundersen wrote:
>> Commit 78551277e4df5: "Input: i8042 - add PNP modaliases" had a bug, where the
>> second call to MODULE_DEVICE_TABLE() overrode the first resulting in not all
>> the modaliases being exposed.
>> 
>> This fixes the problem by including the name of the device_id table in the
>> __mod_*_device_table alias, allowing us to export several device_id tables
>> per module.
>> 
>> Suggested-by: Kay Sievers <kay@vrfy.org>
>> Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
>> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
>> Cc: Rusty Russell <rusty@rustcorp.com.au>
>> ---
>>  include/linux/module.h   |  2 +-
>>  scripts/mod/file2alias.c | 14 +++++++++-----
>>  2 files changed, 10 insertions(+), 6 deletions(-)
>
> Ah, very nice, I've wanted this for a while now, it would make a number
> of different drivers much smaller and simpler to add new device ids to
> (no multiple lists of ids, one for the module loader and one for the
> sub-driver that is in the single file.)

You never asked :(

I've applied, this, but I'm actually amazed this patch works.  C is
weird sometimes.  It changes declarations of the form:

extern const struct pci_device_id __mod_pci_device_id_table
        __attribute__ ((unused, alias("e1000_pci_tbl"));

Into:

extern const struct pci__e1000_pci_tbl_device_id __mod_pci__e1000_pci_tbl_device_id_table
        __attribute__ ((unused, alias("e1000_pci_tbl"));

Now, that's a completely unknown type, but gcc doesn't care because it's
just an extern declaration.  It does insert the alias, which is all we
care about.

We would normally use a special section for this, so it's mainly
historical.  Now we have DEFINE_PCI_DEVICE_TABLE etc, we should
use those to put it in a special section (eg. "pci_ids") and
grab that directly.

Volunteers welcome :)
Rusty.
--
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Rusty Russell Jan. 28, 2014, 1:35 a.m. UTC | #4
Tom Gundersen <teg@jklm.no> writes:
> Commit 78551277e4df5: "Input: i8042 - add PNP modaliases" had a bug, where the
> second call to MODULE_DEVICE_TABLE() overrode the first resulting in not all
> the modaliases being exposed.

No Signed-off-by?

Thanks,
Rusty.

>
> This fixes the problem by including the name of the device_id table in the
> __mod_*_device_table alias, allowing us to export several device_id tables
> per module.
>
> Suggested-by: Kay Sievers <kay@vrfy.org>
> Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> Cc: Rusty Russell <rusty@rustcorp.com.au>
> ---
>  include/linux/module.h   |  2 +-
>  scripts/mod/file2alias.c | 14 +++++++++-----
>  2 files changed, 10 insertions(+), 6 deletions(-)
>
> diff --git a/include/linux/module.h b/include/linux/module.h
> index 15cd6b1..7732d76 100644
> --- a/include/linux/module.h
> +++ b/include/linux/module.h
> @@ -143,7 +143,7 @@ extern const struct gtype##_id __mod_##gtype##_table		\
>  #define MODULE_DESCRIPTION(_description) MODULE_INFO(description, _description)
>  
>  #define MODULE_DEVICE_TABLE(type,name)		\
> -  MODULE_GENERIC_TABLE(type##_device,name)
> +  MODULE_GENERIC_TABLE(type##__##name##_device,name)
>  
>  /* Version of form [<epoch>:]<version>[-<extra-version>].
>     Or for CVS/RCS ID version, everything but the number is stripped.
> diff --git a/scripts/mod/file2alias.c b/scripts/mod/file2alias.c
> index 2370863..6778381 100644
> --- a/scripts/mod/file2alias.c
> +++ b/scripts/mod/file2alias.c
> @@ -42,7 +42,7 @@ typedef unsigned char	__u8;
>  
>  /* This array collects all instances that use the generic do_table */
>  struct devtable {
> -	const char *device_id; /* name of table, __mod_<name>_device_table. */
> +	const char *device_id; /* name of table, __mod_<name>__*_device_table. */
>  	unsigned long id_size;
>  	void *function;
>  };
> @@ -146,7 +146,8 @@ static void device_id_check(const char *modname, const char *device_id,
>  
>  	if (size % id_size || size < id_size) {
>  		fatal("%s: sizeof(struct %s_device_id)=%lu is not a modulo "
> -		      "of the size of section __mod_%s_device_table=%lu.\n"
> +		      "of the size of "
> +		      "section __mod_%s__<identifier>_device_table=%lu.\n"
>  		      "Fix definition of struct %s_device_id "
>  		      "in mod_devicetable.h\n",
>  		      modname, device_id, id_size, device_id, size, device_id);
> @@ -1206,7 +1207,7 @@ void handle_moddevtable(struct module *mod, struct elf_info *info,
>  {
>  	void *symval;
>  	char *zeros = NULL;
> -	const char *name;
> +	const char *name, *identifier;
>  	unsigned int namelen;
>  
>  	/* We're looking for a section relative symbol */
> @@ -1217,7 +1218,7 @@ void handle_moddevtable(struct module *mod, struct elf_info *info,
>  	if (ELF_ST_TYPE(sym->st_info) != STT_OBJECT)
>  		return;
>  
> -	/* All our symbols are of form <prefix>__mod_XXX_device_table. */
> +	/* All our symbols are of form <prefix>__mod_<name>__<identifier>_device_table. */
>  	name = strstr(symname, "__mod_");
>  	if (!name)
>  		return;
> @@ -1227,7 +1228,10 @@ void handle_moddevtable(struct module *mod, struct elf_info *info,
>  		return;
>  	if (strcmp(name + namelen - strlen("_device_table"), "_device_table"))
>  		return;
> -	namelen -= strlen("_device_table");
> +	identifier = strstr(name, "__");
> +	if (!identifier)
> +		return;
> +	namelen = identifier - name;
>  
>  	/* Handle all-NULL symbols allocated into .bss */
>  	if (info->sechdrs[get_secindex(info, sym)].sh_type & SHT_NOBITS) {
> -- 
> 1.8.5.3
--
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Tom Gundersen Jan. 28, 2014, 12:38 p.m. UTC | #5
Hi Rusty,

On Tue, Jan 28, 2014 at 2:35 AM, Rusty Russell <rusty@rustcorp.com.au> wrote:
> Tom Gundersen <teg@jklm.no> writes:
>> Commit 78551277e4df5: "Input: i8042 - add PNP modaliases" had a bug, where the
>> second call to MODULE_DEVICE_TABLE() overrode the first resulting in not all
>> the modaliases being exposed.
>
> No Signed-off-by?

Sorry about that, I'll resend in a minute with the signoff.

Cheers,

Tom
--
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/include/linux/module.h b/include/linux/module.h
index 15cd6b1..7732d76 100644
--- a/include/linux/module.h
+++ b/include/linux/module.h
@@ -143,7 +143,7 @@  extern const struct gtype##_id __mod_##gtype##_table		\
 #define MODULE_DESCRIPTION(_description) MODULE_INFO(description, _description)
 
 #define MODULE_DEVICE_TABLE(type,name)		\
-  MODULE_GENERIC_TABLE(type##_device,name)
+  MODULE_GENERIC_TABLE(type##__##name##_device,name)
 
 /* Version of form [<epoch>:]<version>[-<extra-version>].
    Or for CVS/RCS ID version, everything but the number is stripped.
diff --git a/scripts/mod/file2alias.c b/scripts/mod/file2alias.c
index 2370863..6778381 100644
--- a/scripts/mod/file2alias.c
+++ b/scripts/mod/file2alias.c
@@ -42,7 +42,7 @@  typedef unsigned char	__u8;
 
 /* This array collects all instances that use the generic do_table */
 struct devtable {
-	const char *device_id; /* name of table, __mod_<name>_device_table. */
+	const char *device_id; /* name of table, __mod_<name>__*_device_table. */
 	unsigned long id_size;
 	void *function;
 };
@@ -146,7 +146,8 @@  static void device_id_check(const char *modname, const char *device_id,
 
 	if (size % id_size || size < id_size) {
 		fatal("%s: sizeof(struct %s_device_id)=%lu is not a modulo "
-		      "of the size of section __mod_%s_device_table=%lu.\n"
+		      "of the size of "
+		      "section __mod_%s__<identifier>_device_table=%lu.\n"
 		      "Fix definition of struct %s_device_id "
 		      "in mod_devicetable.h\n",
 		      modname, device_id, id_size, device_id, size, device_id);
@@ -1206,7 +1207,7 @@  void handle_moddevtable(struct module *mod, struct elf_info *info,
 {
 	void *symval;
 	char *zeros = NULL;
-	const char *name;
+	const char *name, *identifier;
 	unsigned int namelen;
 
 	/* We're looking for a section relative symbol */
@@ -1217,7 +1218,7 @@  void handle_moddevtable(struct module *mod, struct elf_info *info,
 	if (ELF_ST_TYPE(sym->st_info) != STT_OBJECT)
 		return;
 
-	/* All our symbols are of form <prefix>__mod_XXX_device_table. */
+	/* All our symbols are of form <prefix>__mod_<name>__<identifier>_device_table. */
 	name = strstr(symname, "__mod_");
 	if (!name)
 		return;
@@ -1227,7 +1228,10 @@  void handle_moddevtable(struct module *mod, struct elf_info *info,
 		return;
 	if (strcmp(name + namelen - strlen("_device_table"), "_device_table"))
 		return;
-	namelen -= strlen("_device_table");
+	identifier = strstr(name, "__");
+	if (!identifier)
+		return;
+	namelen = identifier - name;
 
 	/* Handle all-NULL symbols allocated into .bss */
 	if (info->sechdrs[get_secindex(info, sym)].sh_type & SHT_NOBITS) {