Skip to contentSkip to conversation

swsusp userland interface: additional ioctl

RFC PATCH In review Rafael J. Wysocki 4 participants 7 messages Jan 29, 2006 – Jan 30, 2006

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

RFC PATCH 1/1 swsusp userland interface: additional ioctl

Hi,

The appended patch provides an additional ioctl() for the swsusp userland interface that will allow the suspending and/or resuming utilities to change the kernel log level at the time of suspend/resume (or even set it to 0 if they are going to use some frame buffer graphics etc.).

Comments welcome.

Greetings,
Rafael

 kernel/power/console.c |    5 +++--
 kernel/power/power.h   |    6 +++++-
 kernel/power/user.c    |    5 +++++
 3 files changed, 13 insertions(+), 3 deletions(-)
kernel/power/console.c +3 -2
Index: linux-2.6.16-rc1-mm3/kernel/power/console.c
===================================================================
@@ -9,7 +9,8 @@
 #include <linux/console.h>
 #include "power.h"
 
-static int new_loglevel = 10;
+int suspend_loglevel = DEFAULT_SUSPEND_LOGLEVEL;
+
 static int orig_loglevel;
 #ifdef SUSPEND_CONSOLE
 static int orig_fgconsole, orig_kmsg;
@@ -18,7 +19,7 @@ static int orig_fgconsole, orig_kmsg;
 int pm_prepare_console(void)
 {
 	orig_loglevel = console_loglevel;
-	console_loglevel = new_loglevel;
+	console_loglevel = suspend_loglevel;
 
 #ifdef SUSPEND_CONSOLE
 	acquire_console_sem();
kernel/power/power.h +5 -1
Index: linux-2.6.16-rc1-mm3/kernel/power/power.h
===================================================================
@@ -55,6 +55,9 @@ extern struct pbe *pagedir_nosave;
 extern unsigned long image_size;
 extern int in_suspend;
 extern dev_t swsusp_resume_device;
+extern int suspend_loglevel;
+
+#define DEFAULT_SUSPEND_LOGLEVEL	10
 
 extern asmlinkage int swsusp_arch_suspend(void);
 extern asmlinkage int swsusp_arch_resume(void);
@@ -88,7 +91,8 @@ int snapshot_image_loaded(struct snapsho
 #define SNAPSHOT_GET_SWAP_PAGE		_IOR(SNAPSHOT_IOC_MAGIC, 8, void *)
 #define SNAPSHOT_FREE_SWAP_PAGES	_IO(SNAPSHOT_IOC_MAGIC, 9)
 #define SNAPSHOT_SET_SWAP_FILE		_IOW(SNAPSHOT_IOC_MAGIC, 10, unsigned int)
-#define SNAPSHOT_IOC_MAXNR	10
+#define SNAPSHOT_SET_LOGLEVEL		_IOW(SNAPSHOT_IOC_MAGIC, 11, int)
+#define SNAPSHOT_IOC_MAXNR	11
 
 /**
  *	The bitmap is used for tracing allocated swap pages
kernel/power/user.c +5 -0
Index: linux-2.6.16-rc1-mm3/kernel/power/user.c
===================================================================
@@ -78,6 +78,7 @@ static int snapshot_release(struct inode
 		enable_nonboot_cpus();
 		up(&pm_sem);
 	}
+	suspend_loglevel = DEFAULT_SUSPEND_LOGLEVEL;
 	atomic_inc(&device_available);
 	return 0;
 }
@@ -264,6 +265,10 @@ static int snapshot_ioctl(struct inode *
 		}
 		break;
 
+	case SNAPSHOT_SET_LOGLEVEL:
+		suspend_loglevel = arg <= 10 ? arg : DEFAULT_SUSPEND_LOGLEVEL;
+		break;
+
 	default:
 		error = -ENOTTY;
 
#1 lore raw

Hi!

Quoted 4 lines

The appended patch provides an additional ioctl() for the swsusp userland interface that will allow the suspending and/or resuming utilities to change the kernel log level at the time of suspend/resume (or even set it to 0 if they are going to use some frame buffer graphics etc.).

klogd can already change loglevel... It should be possible to use that to silence suspend, no?

Pavel

Signature · 1 line
Thanks, Sharp!
#2 lore raw

Hi,

Quoted 7 lines · Pavel Machek
Quoted 4 lines

The appended patch provides an additional ioctl() for the swsusp userland interface that will allow the suspending and/or resuming utilities to change the kernel log level at the time of suspend/resume (or even set it to 0 if they are going to use some frame buffer graphics etc.).

klogd can already change loglevel... It should be possible to use that to silence suspend, no?

Well, no. Currently the kernel changes the log level to 10 unconditionally in pm_prepare_console() (that's the reason for the patch, BTW).

Rafael

#3 lore raw

Hi!

Quoted 10 lines
Quoted 7 lines
Quoted 4 lines

The appended patch provides an additional ioctl() for the swsusp userland interface that will allow the suspending and/or resuming utilities to change the kernel log level at the time of suspend/resume (or even set it to 0 if they are going to use some frame buffer graphics etc.).

klogd can already change loglevel... It should be possible to use that to silence suspend, no?

Well, no. Currently the kernel changes the log level to 10 unconditionally in pm_prepare_console() (that's the reason for the patch, BTW).

Can we just kill that log-level setting? It may have been useful in the past, but I do not think it has right to live any more.
Pavel

Signature · 1 line
Thanks, Sharp!
#4 lore raw
Quoted 7 lines · Pavel Machek

Hi!

Quoted 5 lines
Quoted 5 lines
Quoted 4 lines

The appended patch provides an additional ioctl() for the swsusp userland interface that will allow the suspending and/or resuming utilities to change the kernel log level at the time of suspend/resume (or even set it to 0 if they are going to use some frame buffer graphics etc.).

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

klogd can already change loglevel... It should be possible to use that to silence suspend, no?

Well, no. Currently the kernel changes the log level to 10 unconditionally in pm_prepare_console() (that's the reason for the patch, BTW).

Can we just kill that log-level setting? It may have been useful in the past, but I do not think it has right to live any more.

Yes, change that, don't create another interface to the kernel where there is already one.

thanks,

greg k-h

#5 lore raw
In reply to Pavel Machek

Hi,

Quoted 5 lines · Pavel Machek
Quoted 5 lines
Quoted 5 lines
Quoted 4 lines

The appended patch provides an additional ioctl() for the swsusp userland interface that will allow the suspending and/or resuming utilities to change the kernel log level at the time of suspend/resume (or even set it to 0 if they are going to use some frame buffer graphics etc.).

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

klogd can already change loglevel... It should be possible to use that to silence suspend, no?

Well, no. Currently the kernel changes the log level to 10 unconditionally in pm_prepare_console() (that's the reason for the patch, BTW).

Can we just kill that log-level setting? It may have been useful in the past, but I do not think it has right to live any more.

Sure, we can:

 kernel/power/console.c |   12 +-----------
 kernel/power/power.h   |    5 +++++
 2 files changed, 6 insertions(+), 11 deletions(-)
kernel/power/console.c +1 -11
Index: linux-2.6.16-rc1-mm4/kernel/power/console.c
===================================================================
@@ -9,18 +9,11 @@
 #include <linux/console.h>
 #include "power.h"
 
-static int new_loglevel = 10;
-static int orig_loglevel;
 #ifdef SUSPEND_CONSOLE
 static int orig_fgconsole, orig_kmsg;
-#endif
 
 int pm_prepare_console(void)
 {
-	orig_loglevel = console_loglevel;
-	console_loglevel = new_loglevel;
-
-#ifdef SUSPEND_CONSOLE
 	acquire_console_sem();
 
 	orig_fgconsole = fg_console;
@@ -41,18 +34,15 @@ int pm_prepare_console(void)
 	}
 	orig_kmsg = kmsg_redirect;
 	kmsg_redirect = SUSPEND_CONSOLE;
-#endif
 	return 0;
 }
 
 void pm_restore_console(void)
 {
-	console_loglevel = orig_loglevel;
-#ifdef SUSPEND_CONSOLE
 	acquire_console_sem();
 	set_console(orig_fgconsole);
 	release_console_sem();
 	kmsg_redirect = orig_kmsg;
-#endif
 	return;
 }
+#endif
kernel/power/power.h +5 -0
Index: linux-2.6.16-rc1-mm4/kernel/power/power.h
===================================================================
@@ -43,8 +43,13 @@ static struct subsys_attribute _name##_a
 
 extern struct subsystem power_subsys;
 
+#ifdef SUSPEND_CONSOLE
 extern int pm_prepare_console(void);
 extern void pm_restore_console(void);
+#else
+static int pm_prepare_console(void) { return 0; }
+static void pm_restore_console(void) {}
+#endif
 
 /* References to section boundaries */
 extern const void __nosave_begin, __nosave_end;
#6 lore raw

Hi!

Quoted 11 lines
Quoted 11 lines
Quoted 10 lines
Quoted 7 lines
Quoted 4 lines

The appended patch provides an additional ioctl() for the swsusp userland interface that will allow the suspending and/or resuming utilities to change the kernel log level at the time of suspend/resume (or even set it to 0 if they are going to use some frame buffer graphics etc.).

klogd can already change loglevel... It should be possible to use that to silence suspend, no?

Well, no. Currently the kernel changes the log level to 10 unconditionally in pm_prepare_console() (that's the reason for the patch, BTW).

Quoted 6 lines
Quoted 2 lines

Can we just kill that log-level setting? It may have been useful in the past, but I do not think it has right to live any more.

Sure, we can:

ACK and thanks. Patches that remve code are always good.

Pavel

Signature · 1 line
Thanks, Sharp!