Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1373537 > unrolled thread
| Started by | Sergey Senozhatsky <sergey.senozhatsky@gmail.com> |
|---|---|
| First post | 2016-04-07 18:40 +0200 |
| Last post | 2016-04-08 07:50 +0200 |
| Articles | 4 — 3 participants |
Back to article view | Back to linux.kernel
This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by
below is the oldest one visible, not the original post.
[PATCH v11 3/3] printk: make printk.synchronous param rw Sergey Senozhatsky <sergey.senozhatsky@gmail.com> - 2016-04-07 18:40 +0200
Re: [PATCH v11 3/3] printk: make printk.synchronous param rw Pan Xinhui <xinhui@linux.vnet.ibm.com> - 2016-04-08 06:10 +0200
Re: [PATCH v11 3/3] printk: make printk.synchronous param rw Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com> - 2016-04-08 07:30 +0200
Re: [PATCH v11 3/3] printk: make printk.synchronous param rw Pan Xinhui <xinhui@linux.vnet.ibm.com> - 2016-04-08 07:50 +0200
| From | Sergey Senozhatsky <sergey.senozhatsky@gmail.com> |
|---|---|
| Date | 2016-04-07 18:40 +0200 |
| Subject | [PATCH v11 3/3] printk: make printk.synchronous param rw |
| Message-ID | <rll8m-6JT-17@gated-at.bofh.it> |
Change `synchronous' printk param to be RW, so user space
can change printk mode back and forth to/from sync mode
(which is considered to be more reliable).
Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
---
kernel/printk/printk.c | 63 ++++++++++++++++++++++++++++++++++++++++++--------
1 file changed, 53 insertions(+), 10 deletions(-)
diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
index 89f5441..5ae6f73 100644
--- a/kernel/printk/printk.c
+++ b/kernel/printk/printk.c
@@ -288,9 +288,6 @@ static u32 log_buf_len = __LOG_BUF_LEN;
/* Control whether printing to console must be synchronous. */
static bool __read_mostly printk_sync = true;
-module_param_named(synchronous, printk_sync, bool, S_IRUGO);
-MODULE_PARM_DESC(synchronous, "make printing to console synchronous");
-
/* Printing kthread for async printk */
static struct task_struct *printk_kthread;
/* When `true' printing thread has messages to print */
@@ -1785,7 +1782,7 @@ asmlinkage int vprintk_emit(int facility, int level,
* operate in sync mode once panic() occurred.
*/
if (console_loglevel != CONSOLE_LOGLEVEL_MOTORMOUTH &&
- printk_kthread) {
+ !printk_sync && printk_kthread) {
/* Offload printing to a schedulable context. */
printk_kthread_need_flush_console = true;
wake_up_process(printk_kthread);
@@ -2757,6 +2754,16 @@ static int __init printk_late_init(void)
late_initcall(printk_late_init);
#if defined CONFIG_PRINTK
+/*
+ * kernel_param_ops.set is called from two places:
+ * - from parse_args()->printk_sync_set(), when we can't kthread_run(), so
+ * we just set the param value. The actual initalization happens later,
+ * from late_initcall().
+ *
+ * - from user space via sysfs knob; we can kthread_run() there (if needed).
+ */
+static bool printk_initcall_done;
+
static int printk_kthread_func(void *data)
{
while (1) {
@@ -2780,11 +2787,7 @@ static int printk_kthread_func(void *data)
return 0;
}
-/*
- * Init async printk via late_initcall, after core/arch/device/etc.
- * initialization.
- */
-static int __init init_printk_kthread(void)
+static int __init_printk_kthread(void)
{
struct task_struct *thread;
struct sched_param param = {
@@ -2794,6 +2797,9 @@ static int __init init_printk_kthread(void)
if (printk_sync)
return 0;
+ if (printk_kthread)
+ return 0;
+
thread = kthread_run(printk_kthread_func, NULL, "printk");
if (IS_ERR(thread)) {
pr_err("printk: unable to create printing thread\n");
@@ -2805,6 +2811,43 @@ static int __init init_printk_kthread(void)
printk_kthread = thread;
return 0;
}
+
+static int printk_sync_set(const char *val, const struct kernel_param *kp)
+{
+ static DEFINE_MUTEX(printk_sync_lock);
+ int ret;
+
+ mutex_lock(&printk_sync_lock);
+ ret = param_set_bool(val, kp);
+ if (ret) {
+ mutex_unlock(&printk_sync_lock);
+ return ret;
+ }
+
+ if (printk_initcall_done)
+ ret = __init_printk_kthread();
+ mutex_unlock(&printk_sync_lock);
+ return ret;
+}
+
+static const struct kernel_param_ops param_ops_printk_sync = {
+ .set = printk_sync_set,
+ .get = param_get_bool,
+};
+
+module_param_cb(synchronous, ¶m_ops_printk_sync, &printk_sync,
+ S_IRUGO | S_IWUSR);
+MODULE_PARM_DESC(synchronous, "make printing to console synchronous");
+
+/*
+ * Init async printk via late_initcall, after core/arch/device/etc.
+ * initialization.
+ */
+static __init int init_printk_kthread(void)
+{
+ printk_initcall_done = true;
+ return __init_printk_kthread();
+}
late_initcall(init_printk_kthread);
/*
@@ -2820,7 +2863,7 @@ static void wake_up_klogd_work_func(struct irq_work *irq_work)
int pending = __this_cpu_xchg(printk_pending, 0);
if (pending & PRINTK_PENDING_OUTPUT) {
- if (printk_kthread) {
+ if (!printk_sync && printk_kthread) {
wake_up_process(printk_kthread);
} else {
/*
--
2.8.0
[toc] | [next] | [standalone]
| From | Pan Xinhui <xinhui@linux.vnet.ibm.com> |
|---|---|
| Date | 2016-04-08 06:10 +0200 |
| Message-ID | <rlvU6-6xI-9@gated-at.bofh.it> |
| In reply to | #1373537 |
On 2016年04月08日 01:31, Sergey Senozhatsky wrote:
> Change `synchronous' printk param to be RW, so user space
> can change printk mode back and forth to/from sync mode
> (which is considered to be more reliable).
>
> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
> ---
> kernel/printk/printk.c | 63 ++++++++++++++++++++++++++++++++++++++++++--------
> 1 file changed, 53 insertions(+), 10 deletions(-)
>
> diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
> index 89f5441..5ae6f73 100644
> --- a/kernel/printk/printk.c
> +++ b/kernel/printk/printk.c
> @@ -288,9 +288,6 @@ static u32 log_buf_len = __LOG_BUF_LEN;
>
> /* Control whether printing to console must be synchronous. */
> static bool __read_mostly printk_sync = true;
> -module_param_named(synchronous, printk_sync, bool, S_IRUGO);
> -MODULE_PARM_DESC(synchronous, "make printing to console synchronous");
> -
> /* Printing kthread for async printk */
> static struct task_struct *printk_kthread;
> /* When `true' printing thread has messages to print */
> @@ -1785,7 +1782,7 @@ asmlinkage int vprintk_emit(int facility, int level,
> * operate in sync mode once panic() occurred.
> */
> if (console_loglevel != CONSOLE_LOGLEVEL_MOTORMOUTH &&
> - printk_kthread) {
> + !printk_sync && printk_kthread) {
> /* Offload printing to a schedulable context. */
> printk_kthread_need_flush_console = true;
> wake_up_process(printk_kthread);
> @@ -2757,6 +2754,16 @@ static int __init printk_late_init(void)
> late_initcall(printk_late_init);
>
> #if defined CONFIG_PRINTK
> +/*
> + * kernel_param_ops.set is called from two places:
> + * - from parse_args()->printk_sync_set(), when we can't kthread_run(), so
> + * we just set the param value. The actual initalization happens later,
> + * from late_initcall().
> + *
> + * - from user space via sysfs knob; we can kthread_run() there (if needed).
> + */
> +static bool printk_initcall_done;
> +
> static int printk_kthread_func(void *data)
> {
> while (1) {
> @@ -2780,11 +2787,7 @@ static int printk_kthread_func(void *data)
> return 0;
> }
>
> -/*
> - * Init async printk via late_initcall, after core/arch/device/etc.
> - * initialization.
> - */
> -static int __init init_printk_kthread(void)
> +static int __init_printk_kthread(void)
> {
> struct task_struct *thread;
> struct sched_param param = {
> @@ -2794,6 +2797,9 @@ static int __init init_printk_kthread(void)
> if (printk_sync)
> return 0;
>
> + if (printk_kthread)
> + return 0;
> +
> thread = kthread_run(printk_kthread_func, NULL, "printk");
> if (IS_ERR(thread)) {
> pr_err("printk: unable to create printing thread\n");
> @@ -2805,6 +2811,43 @@ static int __init init_printk_kthread(void)
> printk_kthread = thread;
> return 0;
> }
> +
> +static int printk_sync_set(const char *val, const struct kernel_param *kp)
> +{
> + static DEFINE_MUTEX(printk_sync_lock);
> + int ret;
> +
> + mutex_lock(&printk_sync_lock);
> + ret = param_set_bool(val, kp);
> + if (ret) {
> + mutex_unlock(&printk_sync_lock);
> + return ret;
> + }
> +
> + if (printk_initcall_done)
> + ret = __init_printk_kthread();
> + mutex_unlock(&printk_sync_lock);
> + return ret;
> +}
> +
> +static const struct kernel_param_ops param_ops_printk_sync = {
> + .set = printk_sync_set,
> + .get = param_get_bool,
> +};
> +
> +module_param_cb(synchronous, ¶m_ops_printk_sync, &printk_sync,
> + S_IRUGO | S_IWUSR);
> +MODULE_PARM_DESC(synchronous, "make printing to console synchronous");
> +
> +/*
> + * Init async printk via late_initcall, after core/arch/device/etc.
> + * initialization.
> + */
> +static __init int init_printk_kthread(void)
> +{
> + printk_initcall_done = true;
> + return __init_printk_kthread();
hello,
One confusion, Why not use a lock to protect __init_printk_kthread from parallel call? Otherwise I think there is a race.
But for simplicity, maybe you could write codes as below.
+ int ret = __init_printk_kthread();
+ printk_initcall_done = true;
+ return ret;
In my opinion, using a lock is better.
thanks
xinhui
> +}
> late_initcall(init_printk_kthread);
>
> /*
> @@ -2820,7 +2863,7 @@ static void wake_up_klogd_work_func(struct irq_work *irq_work)
> int pending = __this_cpu_xchg(printk_pending, 0);
>
> if (pending & PRINTK_PENDING_OUTPUT) {
> - if (printk_kthread) {
> + if (!printk_sync && printk_kthread) {
> wake_up_process(printk_kthread);
> } else {
> /*
>
[toc] | [prev] | [next] | [standalone]
| From | Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com> |
|---|---|
| Date | 2016-04-08 07:30 +0200 |
| Message-ID | <rlx9w-7sb-13@gated-at.bofh.it> |
| In reply to | #1373952 |
On (04/08/16 12:04), Pan Xinhui wrote:
[..]
> > +/*
> > + * Init async printk via late_initcall, after core/arch/device/etc.
> > + * initialization.
> > + */
> > +static __init int init_printk_kthread(void)
> > +{
> > + printk_initcall_done = true;
> > + return __init_printk_kthread();
> hello,
>
> One confusion, Why not use a lock to protect __init_printk_kthread from parallel call? Otherwise I think there is a race.
> But for simplicity, maybe you could write codes as below.
>
> + int ret = __init_printk_kthread();
> + printk_initcall_done = true;
> + return ret;
>
> In my opinion, using a lock is better.
Hello,
I though about this, but isn't late_initcall() happening before kernel
starts /sbin/init? who can race with
late_initcall() -> init_printk_kthread() -> __init_printk_kthread()?
looking at
static int __ref kernel_init(void *unused)
{
int ret;
kernel_init_freeable();
/* need to finish all async __init code before freeing the memory */
async_synchronize_full();
free_initmem();
..
if (!try_to_run_init_process("/sbin/init") ||
!try_to_run_init_process("/etc/init") ||
!try_to_run_init_process("/bin/init") ||
!try_to_run_init_process("/bin/sh"))
return 0;
__init (and init_printk_kthread is __init) is finished and freed by the
time kernel try_to_run_init_process. isn't it?
sysfs knob -> __init_printk_kthread() is protected by printk_sync_lock
mutex, obviously there can be parallel calls from user space.
-ss
[toc] | [prev] | [next] | [standalone]
| From | Pan Xinhui <xinhui@linux.vnet.ibm.com> |
|---|---|
| Date | 2016-04-08 07:50 +0200 |
| Message-ID | <rlxsS-7C5-11@gated-at.bofh.it> |
| In reply to | #1373990 |
On 2016年04月08日 13:29, Sergey Senozhatsky wrote:
> On (04/08/16 12:04), Pan Xinhui wrote:
> [..]
>>> +/*
>>> + * Init async printk via late_initcall, after core/arch/device/etc.
>>> + * initialization.
>>> + */
>>> +static __init int init_printk_kthread(void)
>>> +{
>>> + printk_initcall_done = true;
>>> + return __init_printk_kthread();
>> hello,
>>
>> One confusion, Why not use a lock to protect __init_printk_kthread from parallel call? Otherwise I think there is a race.
>> But for simplicity, maybe you could write codes as below.
>>
>> + int ret = __init_printk_kthread();
>> + printk_initcall_done = true;
>> + return ret;
>>
>> In my opinion, using a lock is better.
>
> Hello,
>
> I though about this, but isn't late_initcall() happening before kernel
> starts /sbin/init? who can race with
>
> late_initcall() -> init_printk_kthread() -> __init_printk_kthread()?
>
yep, you are right. I made a mistake.
> looking at
>
> static int __ref kernel_init(void *unused)
> {
> int ret;
>
> kernel_init_freeable();
> /* need to finish all async __init code before freeing the memory */
> async_synchronize_full();
> free_initmem();
> ..
>
> if (!try_to_run_init_process("/sbin/init") ||
> !try_to_run_init_process("/etc/init") ||
> !try_to_run_init_process("/bin/init") ||
> !try_to_run_init_process("/bin/sh"))
> return 0;
>
> __init (and init_printk_kthread is __init) is finished and freed by the
> time kernel try_to_run_init_process. isn't it?
>
>
good explanation. thanks
> sysfs knob -> __init_printk_kthread() is protected by printk_sync_lock
> mutex, obviously there can be parallel calls from user space.
>
> -ss
>
[toc] | [prev] | [standalone]
Back to top | Article view | linux.kernel
csiph-web