swich attribute open() to simple stat()

swich attribute open() to simple stat()

[Kay Sievers]

[-- Attachment #1: Type: text/plain, Size: 274 bytes --]

Here we switch to a simple stat() to recognize the file creation. This
will make it possible to wait for files, which are unreadable.

I've switched the timout back to 5 seconds cause the our /class/vc errors
seems like a bug in the vc layer not a sysfs delay.

Thanks,
Kay

[-- Attachment #2: udev-stat-instead-open.patch --]
[-- Type: text/plain, Size: 1451 bytes --]

===== wait_for_sysfs.c 1.18 vs edited =====
--- 1.18/wait_for_sysfs.c	2004-10-14 16:14:00 +02:00
+++ edited/wait_for_sysfs.c	2004-10-15 19:32:38 +02:00
@@ -47,7 +47,7 @@ void log_message(int level, const char *
 }
 #endif
 
-#define WAIT_MAX_SECONDS		10
+#define WAIT_MAX_SECONDS		5
 #define WAIT_LOOP_PER_SECOND		20
 
 /* wait for specific file to show up, normally the "dev"-file */
@@ -60,7 +60,7 @@ static int wait_for_class_device_attribu
 		{ .subsystem = "net",		.file = "ifindex" },
 		{ .subsystem = "scsi_host",	.file = "unique_id" },
 		{ .subsystem = "scsi_device",	.file = NULL },
-		{ .subsystem = "pcmcia_socket",	.file = NULL }, /* all files are unreadable in empty slot :( */
+		{ .subsystem = "pcmcia_socket",	.file = "card_type" },
 		{ .subsystem = "usb_host",	.file = NULL },
 		{ .subsystem = "bluetooth",	.file = "address" },
 		{ .subsystem = "firmware",	.file = "data" },
@@ -86,11 +86,14 @@ static int wait_for_class_device_attribu
 			break;
 		}
 	}
-	dbg("looking at class '%s' for specific file '%s'", class_dev->classname, file);
+
+	dbg("looking at class '%s' for specific file '%s'", class_dev->classname, class_dev->path);
 
 	loop = WAIT_MAX_SECONDS * WAIT_LOOP_PER_SECOND;
 	while (--loop) {
-		if (sysfs_get_classdev_attr(class_dev, file) != NULL) {
+		struct stat stats;
+
+		if (stat(class_dev->path, &stats) == 0) {
 			dbg("class '%s' specific file '%s' found", class_dev->classname, file);
 			return 0;
 		}

Re: swich attribute open() to simple stat()

[Greg KH]

On Fri, Oct 15, 2004 at 07:36:19PM +0200, Kay Sievers wrote:
> Here we switch to a simple stat() to recognize the file creation. This
> will make it possible to wait for files, which are unreadable.
> 
> I've switched the timout back to 5 seconds cause the our /class/vc errors
> seems like a bug in the vc layer not a sysfs delay.

Applied, thanks.

greg k-h


-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: swich attribute open() to simple stat()

[Greg KH]

On Fri, Oct 15, 2004 at 07:36:19PM +0200, Kay Sievers wrote:
> Here we switch to a simple stat() to recognize the file creation. This
> will make it possible to wait for files, which are unreadable.
> 
> I've switched the timout back to 5 seconds cause the our /class/vc errors
> seems like a bug in the vc layer not a sysfs delay.
> 
> Thanks,
> Kay

> === wait_for_sysfs.c 1.18 vs edited ==> --- 1.18/wait_for_sysfs.c	2004-10-14 16:14:00 +02:00
> +++ edited/wait_for_sysfs.c	2004-10-15 19:32:38 +02:00
> @@ -47,7 +47,7 @@ void log_message(int level, const char *
>  }
>  #endif
>  
> -#define WAIT_MAX_SECONDS		10
> +#define WAIT_MAX_SECONDS		5
>  #define WAIT_LOOP_PER_SECOND		20
>  
>  /* wait for specific file to show up, normally the "dev"-file */
> @@ -60,7 +60,7 @@ static int wait_for_class_device_attribu
>  		{ .subsystem = "net",		.file = "ifindex" },
>  		{ .subsystem = "scsi_host",	.file = "unique_id" },
>  		{ .subsystem = "scsi_device",	.file = NULL },
> -		{ .subsystem = "pcmcia_socket",	.file = NULL }, /* all files are unreadable in empty slot :( */
> +		{ .subsystem = "pcmcia_socket",	.file = "card_type" },
>  		{ .subsystem = "usb_host",	.file = NULL },
>  		{ .subsystem = "bluetooth",	.file = "address" },
>  		{ .subsystem = "firmware",	.file = "data" },
> @@ -86,11 +86,14 @@ static int wait_for_class_device_attribu
>  			break;
>  		}
>  	}
> -	dbg("looking at class '%s' for specific file '%s'", class_dev->classname, file);
> +
> +	dbg("looking at class '%s' for specific file '%s'", class_dev->classname, class_dev->path);
>  
>  	loop = WAIT_MAX_SECONDS * WAIT_LOOP_PER_SECOND;
>  	while (--loop) {
> -		if (sysfs_get_classdev_attr(class_dev, file) != NULL) {
> +		struct stat stats;
> +
> +		if (stat(class_dev->path, &stats) = 0) {

Wait, this patch stops us from looking for "file" which had been set to
the file we wanted to look at, right?  Now we are just verifying that
the sysfs _directory_ for the class device is present.

While this is a good check (and we should do this to verify that the
device hasn't gone away, like in the vcs issue), I don't think this is
what you wanted to do, right?

thanks,

greg k-h


-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: swich attribute open() to simple stat()

[Greg KH]

On Fri, Oct 15, 2004 at 01:55:21PM -0700, Greg KH wrote:
> 
> While this is a good check (and we should do this to verify that the
> device hasn't gone away, like in the vcs issue), I don't think this is
> what you wanted to do, right?

Ok, I've fixed this up in the bk tree.  A new udev release will be
soon...

thanks,

greg k-h


-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: swich attribute open() to simple stat()

[Kay Sievers]

On Fri, Oct 15, 2004 at 01:55:21PM -0700, Greg KH wrote:
> On Fri, Oct 15, 2004 at 07:36:19PM +0200, Kay Sievers wrote:
> > Here we switch to a simple stat() to recognize the file creation. This
> > will make it possible to wait for files, which are unreadable.
> > 
> > I've switched the timout back to 5 seconds cause the our /class/vc errors
> > seems like a bug in the vc layer not a sysfs delay.

> > -		if (sysfs_get_classdev_attr(class_dev, file) != NULL) {
> > +		struct stat stats;
> > +
> > +		if (stat(class_dev->path, &stats) = 0) {
> 
> Wait, this patch stops us from looking for "file" which had been set to
> the file we wanted to look at, right?  Now we are just verifying that
> the sysfs _directory_ for the class device is present.
> 
> While this is a good check (and we should do this to verify that the
> device hasn't gone away, like in the vcs issue), I don't think this is
> what you wanted to do, right?

Yeah, you're right, sorry.

Kay


-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel