* [PATCH 1/2] printk: Make continuation flags from /dev/kmsg useful again
[not found] <20191119170632.52119-1-james.byrne@origamienergy.com>
@ 2019-11-19 17:08 ` James Byrne
2019-11-19 17:08 ` [PATCH 2/2] printk: Support message continuation from /dev/kmsg James Byrne
1 sibling, 0 replies; 2+ messages in thread
From: James Byrne @ 2019-11-19 17:08 UTC (permalink / raw)
To: LKML; +Cc: James Byrne, Petr Mladek, Sergey Senozhatsky, Steven Rostedt
Commit 5aa068ea4082 ("printk: remove games with previous record flags")
abolished the practice of setting the log flag to 'c' for the first
continuation line and '+' for subsequent lines. Since this change all
continuation lines are flagged with 'c' and '+' is never used.
From the point of view of a consumer of /dev/kmsg, the behaviour now
is not very useful because you cannot join lines together based on the
'c' flag since you do not know whether a line starting with 'c' was
actually a continuation of the previous line or not.
This commit changes the flag field emitted in /dev/kmsg to expose the
log buffer flags in a more useful way:
- If LOG_NEWLINE=1 and LOG_CONT=0 the flag will be '-', meaning this is
a single self-contained line.
- If LOG_NEWLINE=0 and LOG_CONT=0 the flag will be 'c', meaning that
this is potentially the start of a sequence of continuation lines.
- If LOG_NEWLINE=0 and LOG_CONT=1 the flag will be '+', meaning that
this is the middle of a sequence of continuation lines.
- If LOG_NEWLINE=1 and LOG_CONT=1 the flag will be '*', meaning that
this is the end of a sequence of continuation lines.
This allows a consumer to concatenate continuations in a straightforward
manner.
Signed-off-by: James Byrne <james.byrne@origamienergy.com>
---
Documentation/ABI/testing/dev-kmsg | 14 ++++++++------
kernel/printk/printk.c | 19 +++++++++++++++++--
2 files changed, 25 insertions(+), 8 deletions(-)
diff --git a/Documentation/ABI/testing/dev-kmsg b/Documentation/ABI/testing/dev-kmsg
index f307506eb54c..6326deeaf5e3 100644
--- a/Documentation/ABI/testing/dev-kmsg
+++ b/Documentation/ABI/testing/dev-kmsg
@@ -90,12 +90,14 @@ Description: The /dev/kmsg character device node provides userspace access
+sound:card0 - subsystem:devname
The flags field carries '-' by default. A 'c' indicates a
- fragment of a line. Note, that these hints about continuation
- lines are not necessarily correct, and the stream could be
- interleaved with unrelated messages, but merging the lines in
- the output usually produces better human readable results. A
- similar logic is used internally when messages are printed to
- the console, /proc/kmsg or the syslog() syscall.
+ fragment of a line. Following fragments are flagged with '+'
+ and the final fragment in a sequence is flagged with '*'. Note
+ that these hints about continuation lines are not necessarily
+ correct, and the stream could be interleaved with unrelated
+ messages, but merging the lines in the output usually produces
+ better human readable results. A similar logic is used
+ internally when messages are printed to the console, /proc/kmsg
+ or the syslog() syscall.
By default, kernel tries to avoid fragments by concatenating
when it can and fragments are rare; however, when extended
diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
index ca65327a6de8..a3db7f5e56d9 100644
--- a/kernel/printk/printk.c
+++ b/kernel/printk/printk.c
@@ -715,7 +715,7 @@ static ssize_t msg_print_ext_header(char *buf, size_t size,
struct printk_log *msg, u64 seq)
{
u64 ts_usec = msg->ts_nsec;
- char caller[20];
+ char flag, caller[20];
#ifdef CONFIG_PRINTK_CALLER
u32 id = msg->caller_id;
@@ -727,9 +727,24 @@ static ssize_t msg_print_ext_header(char *buf, size_t size,
do_div(ts_usec, 1000);
+ switch (msg->flags & (LOG_CONT|LOG_NEWLINE)) {
+ case LOG_CONT|LOG_NEWLINE:
+ flag = '*';
+ break;
+ case LOG_CONT:
+ flag = '+';
+ break;
+ case LOG_NEWLINE:
+ flag = '-';
+ break;
+ default:
+ flag = 'c';
+ break;
+ }
+
return scnprintf(buf, size, "%u,%llu,%llu,%c%s;",
(msg->facility << 3) | msg->level, seq, ts_usec,
- msg->flags & LOG_CONT ? 'c' : '-', caller);
+ flag, caller);
}
static ssize_t msg_print_ext_body(char *buf, size_t size,
--
2.24.0
^ permalink raw reply related [flat|nested] 2+ messages in thread
* [PATCH 2/2] printk: Support message continuation from /dev/kmsg
[not found] <20191119170632.52119-1-james.byrne@origamienergy.com>
2019-11-19 17:08 ` [PATCH 1/2] printk: Make continuation flags from /dev/kmsg useful again James Byrne
@ 2019-11-19 17:08 ` James Byrne
1 sibling, 0 replies; 2+ messages in thread
From: James Byrne @ 2019-11-19 17:08 UTC (permalink / raw)
To: LKML; +Cc: James Byrne, Petr Mladek, Sergey Senozhatsky, Steven Rostedt
Since commit 4bcc595ccd80 ("printk: reinstate KERN_CONT for printing
continuation lines") the behaviour of messages written into /dev/kmsg
from user space has changed. Previously if you wrote a message that
did not end with a newline, followed by one ending with a newline, the
second message was treated as a continuation of the first. This is no
longer the case since for a message to be treated as a continuation, an
explicit KERN_CONT is required at the start, and this cannot be used in
messages written via /dev/kmsg.
This commit allows bit 11 of the facility/level number to be used to set
the continuation flag, so you can write two messages that you want to be
joined into /dev/kmsg like this:
<13>This is a continu
<2061>ation message.\n
Signed-off-by: James Byrne <james.byrne@origamienergy.com>
---
Documentation/ABI/testing/dev-kmsg | 6 +++++-
kernel/printk/printk.c | 6 ++++++
2 files changed, 11 insertions(+), 1 deletion(-)
diff --git a/Documentation/ABI/testing/dev-kmsg b/Documentation/ABI/testing/dev-kmsg
index 6326deeaf5e3..793cf22595fe 100644
--- a/Documentation/ABI/testing/dev-kmsg
+++ b/Documentation/ABI/testing/dev-kmsg
@@ -12,7 +12,8 @@ Description: The /dev/kmsg character device node provides userspace access
The logged line can be prefixed with a <N> syslog prefix, which
carries the syslog priority and facility. The single decimal
prefix number is composed of the 3 lowest bits being the syslog
- priority and the next 8 bits the syslog facility number.
+ priority, the next 8 bits the syslog facility number and the
+ next bit a continuation flag.
If no prefix is given, the priority number is the default kernel
log priority and the facility number is set to LOG_USER (1). It
@@ -20,6 +21,9 @@ Description: The /dev/kmsg character device node provides userspace access
facility number LOG_KERN (0), to make sure that the origin of
the messages can always be reliably determined.
+ Setting bit 11 of the prefix number, the continuation flag, is
+ equivalent to prefixing a kernel printk message with KERN_CONT.
+
Accessing the buffer:
Every read() from the opened device node receives one record
of the kernel's printk buffer.
diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
index a3db7f5e56d9..d04353076e92 100644
--- a/kernel/printk/printk.c
+++ b/kernel/printk/printk.c
@@ -451,6 +451,7 @@ static u32 clear_idx;
#define LOG_LEVEL(v) ((v) & 0x07)
#define LOG_FACILITY(v) ((v) >> 3 & 0xff)
+#define LOG_CONT_USER(v) ((v) & 0x800)
/* record buffer */
#define LOG_ALIGN __alignof__(struct printk_log)
@@ -869,6 +870,8 @@ static ssize_t devkmsg_write(struct kiocb *iocb, struct iov_iter *from)
level = LOG_LEVEL(u);
if (LOG_FACILITY(u) != 0)
facility = LOG_FACILITY(u);
+ if (LOG_CONT_USER(u) != 0)
+ facility |= 0x100;
endp++;
len -= endp - line;
line = endp;
@@ -1954,6 +1957,9 @@ int vprintk_store(int facility, int level,
text_len -= 2;
text += 2;
}
+ } else if (facility & 0x100) {
+ lflags |= LOG_CONT;
+ facility &= 0xff;
}
if (level == LOGLEVEL_DEFAULT)
--
2.24.0
^ permalink raw reply related [flat|nested] 2+ messages in thread
end of thread, other threads:[~2019-11-19 17:09 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
[not found] <20191119170632.52119-1-james.byrne@origamienergy.com>
2019-11-19 17:08 ` [PATCH 1/2] printk: Make continuation flags from /dev/kmsg useful again James Byrne
2019-11-19 17:08 ` [PATCH 2/2] printk: Support message continuation from /dev/kmsg James Byrne
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).