console logging detour via printk
in [Kernel]
|
Prev: [PATCH 3/8] PM: suspend_block: Abort task freezing if a suspend_blocker is active.
Next: [PATCH tip/core/urgent 06/10] sched: Fix an RCU warning in print_task()
From: Samo Pogacnik on 1 May 2010 14:50 Dne 01.05.2010 (sob) ob 12:04 +0100 je Alan Cox zapisal(a): > > while i was searching for effective logging of complete console output > > produced by the kernel and user phase of the boot process, it turned out > > that only kernel messages imho get systematically cached and stored into > > log files (if needed). All userspace processes are on their own to use > > syslog, which is fine, but there are also many console messages > > reporting the boot status via init scripts, .... I came across the > > bootlogd daemo, which handles the job of redirecting console output into > > a log file, but i find it problematic to use especialy, when using > > initial ram disk image. > > So you want to patch the kernel because you can't work out how to do this > in userspace ? The distributions seem to have no problem doing this in > user space that I can see. It doesn't seem to be a hard user space > problem, and there are a ton of things you want to do with this sort of > stuff (like network logging) that you can't do in kernel space. The distros have no problem logging complete console output into log files or over the network, because they simply do not do it at least for the initrd part of the boot process (i'd be glad, if i'm wrong). Anyway the proposed mechanism nicely bridges the gap between the kernel boot start and the system logging daemon start-up. It also solves the chicken and egg problem of how to log console if user space console logging facility fails. > > > --- a_linux-2.6.33.3/drivers/char/vt.c > > +++ b_linux-2.6.33.3/drivers/char/vt.c > > @@ -2696,6 +2696,16 @@ static int con_write(struct tty_struct *tty, const unsigned char *buf, int count > > { > > int retval; > > > > +#ifdef CONFIG_VT_CONSOLE_DETOUR > > + if (console_detour) { > > + int idx = vt_console_driver.index - 1; > > + > > + if ((idx >= 0) && (idx == tty->index)) { > > + console_printk_detour(buf, count); > > + return count; > > + } > > + } > > +#endif > > This requires you go around hacking up each device which is not a good > idea and becomes rapidly unmaintainable. Agree, it should have been done within a well defined code volume. > > I suspect what you actually need for such logging might be to write a > very simple tty driver whose write method is implemented as printk. That > works in the general case and doesn't require hacking up the code > everywhere else. Looks to me that some kernel code is welcome:)? > > However given your init stuff can trivially use openpty to set up a logged > console I am not sure I see the point in doing this in kernel in the > first place. > As said above, how to bridge kernel boot start and logging daemon start-up without kernel help, especially when initrd is in the way? imho it would be too complicated. > Alan > -- > To unsubscribe from this list: send the line "unsubscribe linux-embedded" in > the body of a message to majordomo(a)vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html thanks for the reply, with best regards, Samo -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Alan Cox on 1 May 2010 15:40 > The distros have no problem logging complete console output into log > files or over the network, because they simply do not do it at least for > the initrd part of the boot process (i'd be glad, if i'm wrong). I'd have to double check - but its trivial to move the log if so. > > I suspect what you actually need for such logging might be to write a > > very simple tty driver whose write method is implemented as printk. That > > works in the general case and doesn't require hacking up the code > > everywhere else. > > Looks to me that some kernel code is welcome:)? I really don't see the point but if you must do it then doing it as its own driver would at least avoid making a mess in the rest of the kernel, at which point it becomes less of a problem > > However given your init stuff can trivially use openpty to set up a logged > > console I am not sure I see the point in doing this in kernel in the > > first place. > > > > As said above, how to bridge kernel boot start and logging daemon > start-up without kernel help, especially when initrd is in the way? imho > it would be too complicated. Put the logging start up in the initrd, its just a ramdisk its not special in any way at all. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Samo Pogacnik on 1 May 2010 17:10 Dne 01.05.2010 (sob) ob 18:36 +0200 je Samo Pogacnik zapisal(a): > Dne 01.05.2010 (sob) ob 11:00 +0200 je Geert Uytterhoeven zapisal(a): > > On Sat, May 1, 2010 at 00:03, Samo Pogacnik <samo_pogacnik(a)t-2.net> wrote: > > > while i was searching for effective logging of complete console output > > > produced by the kernel and user phase of the boot process, it turned out > > > that only kernel messages imho get systematically cached and stored into > > > log files (if needed). All userspace processes are on their own to use > > > syslog, which is fine, but there are also many console messages > > > reporting the boot status via init scripts, .... I came across the > > > bootlogd daemo, which handles the job of redirecting console output into > > > a log file, but i find it problematic to use especialy, when using > > > initial ram disk image. > > > > > > So in short i came up with an idea to transform console writes into > > > printks at appropriate code place of some console drivers (the patch > > > includes code for VT console and SERIAL_CORE console drivers). Printks > > > eventually reach console device avoiding the patched part of the console > > > drivers. > > > > What about catching /dev/console instead of VT console, SERIAL_CORE > > console, ...? > > Then it works with whatever console= parameter you specify. > > Could not agree more, but that was as close as i was able to detect the > common code and provide something that actually works. Maybe this is > already enough to cover all boot consoles? Silly me, i managed to miss the common console write method. I'll provide the change, so there is not going to be any specifics for different console types anymore. > > > > > Gr{oetje,eeting}s, > > > > Geert > > > > -- > > Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert(a)linux-m68k.org > > > > In personal conversations with technical people, I call myself a hacker. But > > when I'm talking to journalists I just say "programmer" or something like that. > > -- Linus Torvalds > > regards, Samo > > -- > To unsubscribe from this list: send the line "unsubscribe linux-embedded" in > the body of a message to majordomo(a)vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Samo Pogacnik on 1 May 2010 18:30 Dne 30.04.2010 (pet) ob 15:45 -0700 je Randy Dunlap zapisal(a): > On Sat, 01 May 2010 00:03:00 +0200 Samo Pogacnik wrote: > > > Hi, > > > diff --git a_linux-2.6.33.3/drivers/char/Kconfig b_linux-2.6.33.3/drivers/char/Kconfig > > index e023682..b5d0909 100644 > > --- a_linux-2.6.33.3/drivers/char/Kconfig > > +++ b_linux-2.6.33.3/drivers/char/Kconfig > > @@ -66,6 +66,23 @@ config VT_CONSOLE > > > > If unsure, say Y. > > > > +config VT_CONSOLE_DETOUR > > + bool "Support for VT console detour via printk" > > + depends on VT_CONSOLE > > + default n > > + ---help--- > > + If you do say Y here, the support for writing console messages via > > If you say Y here, > > > + printk is included into VT console code. > > + > > + The feature is usefull to catch all console log. In order to use this > > useful log messages. > > > + feature, you should specify kernel command line option "detour" or write a > > + positive number into /proc/sys/kernel/console_detour. You can disable > > + the feature on-line by writing zero into the proc file. By writing a > > + negative value into the proc file, the feature is disabled permanently > > + (until next boot). > > + > > + If unsure, say N. > > + > > config HW_CONSOLE > > bool > > depends on VT && !S390 && !UML > > > diff --git a_linux-2.6.33.3/drivers/serial/Kconfig b_linux-2.6.33.3/drivers/serial/Kconfig > > index 9ff47db..20acfab 100644 > > --- a_linux-2.6.33.3/drivers/serial/Kconfig > > +++ b_linux-2.6.33.3/drivers/serial/Kconfig > > @@ -1031,6 +1031,23 @@ config SERIAL_CORE > > config SERIAL_CORE_CONSOLE > > bool > > > > +config SERIAL_CORE_CONSOLE_DETOUR > > + bool "Support for serial console detour via printk" > > + depends on SERIAL_CORE_CONSOLE > > + default n > > + ---help--- > > + If you do say Y here, the support for writing console messages via > > If you say Y here, > > > + printk is included into serial console code. > > + > > + The feature is usefull to catch all console log. In order to use this > > useful log messages. > > > + feature, you should specify kernel command line option "detour" or write a > > + positive number into /proc/sys/kernel/console_detour. You can disable > > + the feature on-line by writing zero into the proc file. By writing a > > + negative value into the proc file, the feature is disabled permanently > > + (until next boot). > > + > > + If unsure, say N. > > + > > The kernel command line option needs to be added to Documentation/kernel-parameters.txt > also, please. > > > config CONSOLE_POLL > > bool > > > > diff --git a_linux-2.6.33.3/include/linux/console.h b_linux-2.6.33.3/include/linux/console.h > > index dcca533..bc88030 100644 > > --- a_linux-2.6.33.3/include/linux/console.h > > +++ b_linux-2.6.33.3/include/linux/console.h > > @@ -108,6 +108,12 @@ struct console { > > struct console *next; > > }; > > > > +extern int console_detour; > > +extern void console_printk_detour(const unsigned char *, int); > > Please include parameter names in function prototype(s). > > > + > > +struct ctl_table; > > +int detour_sysctl_handler(struct ctl_table *, int, void __user *, size_t *, loff_t *); > > ditto > > > + > > extern int console_set_on_cmdline; > > > > extern int add_preferred_console(char *name, int idx, char *options); > > > Looks interesting/useful to me. Thanks. > > --- > ~Randy > *** Remember to use Documentation/SubmitChecklist when testing your code *** Hi, Here is the updated patch providing your hints and generalized for any console type. regards, Samo --- Signed-off-by: Samo Pogacnik <samo_pogacnik(a)t-2.net> diff --git a_linux-2.6.33.3/Documentation/kernel-parameters.txt b_linux-2.6.33.3/Documentation/kernel-parameters.txt index e2c7487..ab0072c 100644 --- a_linux-2.6.33.3/Documentation/kernel-parameters.txt +++ b_linux-2.6.33.3/Documentation/kernel-parameters.txt @@ -628,6 +628,8 @@ and is between 256 and 4096 characters. It is defined in the file Defaults to the default architecture's huge page size if not specified. + detour [KNL] Enable console logging detour via printk. + dhash_entries= [KNL] Set number of hash buckets for dentry cache. diff --git a_linux-2.6.33.3/drivers/char/Kconfig b_linux-2.6.33.3/drivers/char/Kconfig index e023682..43c552e 100644 --- a_linux-2.6.33.3/drivers/char/Kconfig +++ b_linux-2.6.33.3/drivers/char/Kconfig @@ -88,6 +88,22 @@ config VT_HW_CONSOLE_BINDING information. For framebuffer console users, please refer to <file:Documentation/fb/fbcon.txt>. +config CONSOLE_DETOUR + bool "Support for console detour via printk" + default n + ---help--- + If you say Y here, the support for writing console messages via + printk is included into the console code. + + The feature is useful to catch all console log messages. + In order to use this feature, you should specify kernel command line + option "detour" or write a positive number into + /proc/sys/kernel/console_detour. You can disable the feature on-line + by writing zero into the proc file. By writing a negative value into + the proc file, the feature is disabled permanently (until next boot). + + If unsure, say N. + config DEVKMEM bool "/dev/kmem virtual device support" default y diff --git a_linux-2.6.33.3/drivers/char/tty_io.c b_linux-2.6.33.3/drivers/char/tty_io.c index 76253cf..f77de34 100644 --- a_linux-2.6.33.3/drivers/char/tty_io.c +++ b_linux-2.6.33.3/drivers/char/tty_io.c @@ -1087,6 +1087,13 @@ ssize_t redirected_tty_write(struct file *file, const char __user *buf, } spin_unlock(&redirect_lock); +#ifdef CONFIG_CONSOLE_DETOUR + if (console_detour) { + console_printk_detour(buf, count); + if (!p) + return count; + } +#endif if (p) { ssize_t res; res = vfs_write(p, buf, count, &p->f_pos); diff --git a_linux-2.6.33.3/include/linux/console.h b_linux-2.6.33.3/include/linux/console.h index dcca533..354a7a8 100644 --- a_linux-2.6.33.3/include/linux/console.h +++ b_linux-2.6.33.3/include/linux/console.h @@ -108,6 +108,13 @@ struct console { struct console *next; }; +extern int console_detour; +extern void console_printk_detour(const unsigned char *buf, int count); + +struct ctl_table; +int detour_sysctl_handler(struct ctl_table *table, int write, + void __user *buffer, size_t *length, loff_t *ppos); + extern int console_set_on_cmdline; extern int add_preferred_console(char *name, int idx, char *options); diff --git a_linux-2.6.33.3/init/main.c b_linux-2.6.33.3/init/main.c index 512ba15..add9e95 100644 --- a_linux-2.6.33.3/init/main.c +++ b_linux-2.6.33.3/init/main.c @@ -25,6 +25,7 @@ #include <linux/bootmem.h> #include <linux/acpi.h> #include <linux/tty.h> +#include <linux/console.h> #include <linux/gfp.h> #include <linux/percpu.h> #include <linux/kmod.h> @@ -249,6 +250,14 @@ static int __init loglevel(char *str) early_param("loglevel", loglevel); +static int __init detour(char *str) +{ + console_detour = 1; + return 0; +} + +early_param("detour", detour); + /* * Unknown boot options get handed to init, unless they look like * unused parameters (modprobe will find them in /proc/cmdline). diff --git a_linux-2.6.33.3/kernel/printk.c b_linux-2.6.33.3/kernel/printk.c index 1751c45..953c6ea 100644 --- a_linux-2.6.33.3/kernel/printk.c +++ b_linux-2.6.33.3/kernel/printk.c @@ -1368,6 +1368,51 @@ static int __init disable_boot_consoles(void) } late_initcall(disable_boot_consoles); +/* + * This option can be enabled with kernel command line option "detour" or + * through a proc file (/proc/sys/kernel/console_detour). It enables console + * logging through printk, if supported by enabled console. + */ +int console_detour; +EXPORT_SYMBOL(console_detour); + +#define DETOUR_STR_SIZE 512 +void console_printk_detour(const unsigned char *buf, int count) +{ + char tmp[DETOUR_STR_SIZE + 1]; + + do { + if (count > DETOUR_STR_SIZE) { + memcpy(tmp, buf, DETOUR_STR_SIZE); + tmp[DETOUR_STR_SIZE] = '\0'; + } else { + memcpy(tmp, buf, count); + tmp[count] = '\0'; + } + count -= DETOUR_STR_SIZE; + printk(KERN_INFO "%s", tmp); + } while (count > 0); +} +EXPORT_SYMBOL(console_printk_detour); + +int detour_sysctl_handler(struct ctl_table *table, int write, + void __user *buffer, size_t *length, loff_t *ppos) +{ + static int disable_forever; + + proc_dointvec(table, write, buffer, length, ppos); + if (write) { + if ((console_detour < 0) || (disable_forever != 0)) { + disable_forever = 1; + console_detour = 0; + return 0; + } + if (console_detour > 1) + console_detour = 1; + } + return 0; +} + #if defined CONFIG_PRINTK /* diff --git a_linux-2.6.33.3/kernel/sysctl.c b_linux-2.6.33.3/kernel/sysctl.c index 8a68b24..ab644cb 100644 --- a_linux-2.6.33.3/kernel/sysctl.c +++ b_linux-2.6.33.3/kernel/sysctl.c @@ -50,6 +50,7 @@ #include <linux/ftrace.h> #include <linux/slow-work.h> #include <linux/perf_event.h> +#include <linux/console.h> #include <asm/uaccess.h> #include <asm/processor.h> @@ -936,6 +937,13 @@ static struct ctl_table kern_table[] = { .proc_handler = proc_dointvec, }, #endif + { + .procname = "console_detour", + .data = &console_detour, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = &detour_sysctl_handler, + }, /* * NOTE: do not add new entries to this table unless you have read * Documentation/sysctl/ctl_unnumbered.txt -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Samo Pogacnik on 1 May 2010 18:50
Dne 01.05.2010 (sob) ob 20:41 +0100 je Alan Cox zapisal(a): > > The distros have no problem logging complete console output into log > > files or over the network, because they simply do not do it at least for > > the initrd part of the boot process (i'd be glad, if i'm wrong). > > I'd have to double check - but its trivial to move the log if so. > > > > I suspect what you actually need for such logging might be to write a > > > very simple tty driver whose write method is implemented as printk. That > > > works in the general case and doesn't require hacking up the code > > > everywhere else. > > > > Looks to me that some kernel code is welcome:)? > > I really don't see the point but if you must do it then doing it as its > own driver would at least avoid making a mess in the rest of the kernel, > at which point it becomes less of a problem > > > > However given your init stuff can trivially use openpty to set up a logged > > > console I am not sure I see the point in doing this in kernel in the > > > first place. > > > > > > > As said above, how to bridge kernel boot start and logging daemon > > start-up without kernel help, especially when initrd is in the way? imho > > it would be too complicated. > > Put the logging start up in the initrd, its just a ramdisk its not > special in any way at all. I understand and can agree with everything you said, however providing an additional mechanism might not hurt. Funny enough by generalizing the patch for any console type, the essential code change is made at the same place where initial console redirection mechanism has been implemented. And again, console redirection can not log anything that has been output to console prior to redirection activation. regards, Samo -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/ |