n_tty cleanup + trace additions
cover n_tty cleanup + trace additions
Greg,
This patchset contains no bug fixes only misc cleanup for the N_TTY line discipline.
Patches 1-4 cleans up code duplication remnants from 3.12's changes. Patch 5 simplifies the snarled logic of how much data is considered 'input available'. I was getting tired of hand checking that logic every time I touched minimum_to_wake.
Patch 6 reduces unnecessary wake ups.
Patch 7 extends the trace usage to aid in identifying and fixing input flow-control bugs (this was used to fix the readline() bug). I also expect to use it to reduce input processing restarts. It's use is disabled by default and only enabled with the file-local define N_TTY_TRACE (similar to TTY_DEBUG_HANGUP in tty_io.c).
Regards,
Peter Hurley (7):
n_tty: Merge .receive_buf() flavors
n_tty: Un-inline slow-path n_tty_receive_char()
n_tty: Un-inline slow-path n_tty_receive_char_closing()
n_tty: Refactor PARMRK doubling checks
n_tty: Refactor input_available_p() by call site
n_tty: Only perform wakeups for waiters
n_tty: trace input/read flow control
drivers/tty/n_tty.c | 159 ++++++++++++++++++++++++++++++++-------------------- 1 file changed, 97 insertions(+), 62 deletions(-)
Signature · 1 line
1.8.1.2
PATCH 1/7 n_tty: Merge .receive_buf() flavors
N_TTY's direct and flow-controlled flavors of the .receive_buf() method are nearly identical; fold together.
--- drivers/tty/n_tty.c | 43 ++++++++++++++++--------------------------- 1 file changed, 16 insertions(+), 27 deletions(-)
drivers/tty/n_tty.c +16 -27
@@ -1675,32 +1675,9 @@ static void __receive_buf(struct tty_struct *tty, const unsigned char *cp, } } -static void n_tty_receive_buf (struct tty_struct *tty,const unsigned char *cp, -char *fp,int count) -{ -int room, n; - -down_read (&tty->termios_rwsem); - -while (1 ) { - room =receive_room (tty); - n =min (count, room); -if (!n) -break ; -__receive_buf (tty, cp, fp, n); - cp += n; -if (fp) - fp += n; - count -= n; - } - - tty->receive_room = room; -n_tty_check_throttle (tty); -up_read (&tty->termios_rwsem); -} - -static int n_tty_receive_buf2 (struct tty_struct *tty,const unsigned char *cp, -char *fp,int count) +static int +n_tty_receive_buf_common (struct tty_struct *tty,const unsigned char *cp, +char *fp,int count,int flow) {struct n_tty_data *ldata = tty->disc_data;int room, n, rcvd =0 ;
@@ -1711,7 +1688,7 @@ static int n_tty_receive_buf2(struct tty_struct *tty, const unsigned char *cp, room =receive_room (tty); n =min (count, room);if (!n) { -if (!room) +if (flow && !room) ldata->no_room =1 ;break ; }
@@ -1730,6 +1707,18 @@ static int n_tty_receive_buf2(struct tty_struct *tty, const unsigned char *cp,return rcvd; } +static void n_tty_receive_buf (struct tty_struct *tty,const unsigned char *cp, +char *fp,int count) +{ +n_tty_receive_buf_common (tty, cp, fp, count,0 ); +} + +static int n_tty_receive_buf2 (struct tty_struct *tty,const unsigned char *cp, +char *fp,int count) +{ +return n_tty_receive_buf_common (tty, cp, fp, count,1 ); +} +int is_ignored (int sig) { return (sigismember(¤t->blocked, sig) ||
Signature · 1 line
1.8.1.2
PATCH 2/7 n_tty: Un-inline slow-path n_tty_receive_char()
In reply to Peter HurleyCommit e60d27c4d8b33ba20896b76b6558f061bc6460ff,n_tty: Factor LNEXT processing from per-char i/o path, mistakenly inlined the non-inline alias, n_tty_receive_char(), for the inline function, n_tty_receive_char_inline().
As n_tty_receive_char() is intended for slow-path char processing only, un-inline it.
--- drivers/tty/n_tty.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/tty/n_tty.c +1 -1
@@ -1418,7 +1418,7 @@ n_tty_receive_char_inline(struct tty_struct *tty, unsigned char c)put_tty_queue (c, ldata); } -static inline void n_tty_receive_char (struct tty_struct *tty,unsigned char c) +static void n_tty_receive_char (struct tty_struct *tty,unsigned char c) {n_tty_receive_char_inline (tty, c); }
Signature · 1 line
1.8.1.2
PATCH 3/7 n_tty: Un-inline slow-path n_tty_receive_char_closing()
In reply to Peter HurleyAlthough n_tty_receive_char_closing() only has one call-site, let the compiler inline instead.
--- drivers/tty/n_tty.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/tty/n_tty.c +1 -2
@@ -1443,8 +1443,7 @@ n_tty_receive_char_fast(struct tty_struct *tty, unsigned char c)put_tty_queue (c, ldata); } -static inline void -n_tty_receive_char_closing (struct tty_struct *tty,unsigned char c) +static void n_tty_receive_char_closing (struct tty_struct *tty,unsigned char c) {if (I_ISTRIP (tty)) c &=0x7f ;
Signature · 1 line
1.8.1.2
PATCH 4/7 n_tty: Refactor PARMRK doubling checks
In reply to Peter HurleyPerform PARMRK doubling checks explicitly; remove ternary idiom and local variable.
--- drivers/tty/n_tty.c | 14 +++++--------- 1 file changed, 5 insertions(+), 9 deletions(-)
drivers/tty/n_tty.c +5 -9
@@ -1258,7 +1258,6 @@ static int n_tty_receive_char_special(struct tty_struct *tty, unsigned char c) {struct n_tty_data *ldata = tty->disc_data; -int parmrk;if (I_IXON (tty)) {if (c ==START_CHAR (tty)) {
@@ -1343,8 +1342,6 @@ n_tty_receive_char_special(struct tty_struct *tty, unsigned char c) }if ((c ==EOL_CHAR (tty)) || (c ==EOL2_CHAR (tty) &&L_IEXTEN (tty))) { - parmrk = (c == (unsigned char )'\377' &&I_PARMRK (tty)) - ?1 :0 ;/* * XXX are EOL_CHAR and EOL2_CHAR echoed?!? */
@@ -1359,7 +1356,7 @@ n_tty_receive_char_special(struct tty_struct *tty, unsigned char c)* XXX does PARMRK doubling happen for * EOL_CHAR and EOL2_CHAR? */ -if (parmrk) +if (c == (unsigned char )'\377' &&I_PARMRK (tty))put_tty_queue (c, ldata); handle_newline:
@@ -1373,7 +1370,6 @@ handle_newline: } } - parmrk = (c == (unsigned char )'\377' &&I_PARMRK (tty)) ?1 :0 ;if (L_ECHO (tty)) {finish_erasing (ldata);if (c =='\n' )
@@ -1387,7 +1383,8 @@ handle_newline:commit_echoes (tty); } -if (parmrk) +/* PARMRK doubling check */ +if (c == (unsigned char )'\377' &&I_PARMRK (tty))put_tty_queue (c, ldata);put_tty_queue (c, ldata);
@@ -1398,7 +1395,6 @@ static inline void n_tty_receive_char_inline(struct tty_struct *tty, unsigned char c) {struct n_tty_data *ldata = tty->disc_data; -int parmrk;if (tty->stopped && !tty->flow_stopped &&I_IXON (tty) &&I_IXANY (tty)) {start_tty (tty);
@@ -1412,8 +1408,8 @@ n_tty_receive_char_inline(struct tty_struct *tty, unsigned char c)echo_char (c, tty);commit_echoes (tty); } - parmrk = (c == (unsigned char )'\377' &&I_PARMRK (tty)) ?1 :0 ; -if (parmrk) +/* PARMRK doubling check */ +if (c == (unsigned char )'\377' &&I_PARMRK (tty))put_tty_queue (c, ldata);put_tty_queue (c, ldata); }
Signature · 1 line
1.8.1.2
PATCH 5/7 n_tty: Refactor input_available_p() by call site
In reply to Peter HurleyDistinguish if caller is n_tty_poll() or n_tty_read(), and set the read/wakeup threshold accordingly.
--- drivers/tty/n_tty.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-)
drivers/tty/n_tty.c +4 -3
@@ -1878,14 +1878,15 @@ static int n_tty_open(struct tty_struct *tty)return 0 ; } -static inline int input_available_p (struct tty_struct *tty,int amt) +static inline int input_available_p (struct tty_struct *tty,int poll) {struct n_tty_data *ldata = tty->disc_data; +int amt = poll && !TIME_CHAR (tty) ?MIN_CHAR (tty) :1 ;if (ldata->icanon && !L_EXTPROC (tty)) {if (ldata->canon_head != ldata->read_tail)return 1 ; - }else if (read_cnt (ldata) >= (amt ? amt :1 )) + }else if (read_cnt (ldata) >= amt)return 1 ;return 0 ;
@@ -2384,7 +2385,7 @@ static unsigned int n_tty_poll(struct tty_struct *tty, struct file *file,poll_wait (file, &tty->read_wait, wait);poll_wait (file, &tty->write_wait, wait); -if (input_available_p (tty,TIME_CHAR (tty) ?0 :MIN_CHAR (tty))) +if (input_available_p (tty,1 )) mask |=POLLIN |POLLRDNORM ;else if (ldata->icanon)n_tty_set_room (tty);
Signature · 1 line
1.8.1.2
Re: [PATCH tty-next 5/7] n_tty: Refactor input_available_p() by call site
Quoted 2 lines · On Fri, 22 Nov 2013 10:59:23 -0500 Peter Hurley <[email protected]>…
Distinguish if caller is
n_tty_poll()orn_tty_read(), and set the read/wakeup threshold accordingly.
That looks very wrong to me. The poll() and read() wakeup should always match. In addition if MIN_CHAR is set without a timeout then it should not wake up until MIN_CHAR characters are present. The MIN_CHAR feature is used by various (these days obscure) protocols to optimise block transfer rates.
Changing this is also a userspace visible API change.
So NAK.
The underlying intention of the system (and SYS5.4) is that you can do
MIN_CHAR = 64
alarm(some_timeout)
while(1) {
poll();
events
read(chunksized blocks)
}while it's hardly as critical with modern fast hardware its still the API and necessary to avoid SWS type behaviours in the queue turning it from block/syscall to byte I/O.
Alan
Re: [PATCH tty-next 5/7] n_tty: Refactor input_available_p() by call site
In reply to Peter HurleyQuoted 4 lines · On Fri, 22 Nov 2013 10:59:23 -0500 Peter Hurley <[email protected]>…
Distinguish if caller is
n_tty_poll()orn_tty_read(), and set the read/wakeup threshold accordingly.Signed-off-by Peter Hurley <[email protected]>
Doh ignore previous.. yes that patch is right. I should pay attention to 0/7 8)
Re: [PATCH tty-next 5/7] n_tty: Refactor input_available_p() by call site
Quoted 10 lines · One Thousand Gnomes
Quoted 4 lines · On Fri, 22 Nov 2013 10:59:23 -0500 Peter Hurley <[email protected]>…
Distinguish if caller is
n_tty_poll()orn_tty_read(), and set the read/wakeup threshold accordingly.Signed-off-by Peter Hurley <[email protected]>Doh ignore previous.. yes that patch is right. I should pay attention to 0/7 8)
Maybe I should add more to the commit message from the cover letter description ?
Re: [PATCH tty-next 5/7] n_tty: Refactor input_available_p() by call site
Quoted 7 lines · On Sat, 23 Nov 2013 21:01:56 -0500 Peter Hurley <[email protected]>…
Quoted 6 lines · One Thousand Gnomes
Quoted 3 lines · On Fri, 22 Nov 2013 10:59:23 -0500 Peter Hurley <[email protected]>…
Distinguish if caller is
n_tty_poll()orn_tty_read(), and set the read/wakeup threshold accordingly.
Quoted 7 lines
Quoted 4 lines
Quoted 1 line
Signed-off-by Peter Hurley <[email protected]>Doh ignore previous.. yes that patch is right. I should pay attention to 0/7 8)
Maybe I should add more to the commit message from the cover letter description ?
Maybe I should just read the code more carefully 8).
Alan
PATCH 6/7 n_tty: Only perform wakeups for waiters
In reply to Peter HurleyOnly wakeup the _waiting_ reader, polls and/or writer(s).
--- drivers/tty/n_tty.c | 18 ++++++++++++------ 1 file changed, 12 insertions(+), 6 deletions(-)
drivers/tty/n_tty.c +12 -6
@@ -275,7 +275,8 @@ static void n_tty_check_unthrottle(struct tty_struct *tty)return ;n_tty_set_room (tty);n_tty_write_wakeup (tty->link); -wake_up_interruptible_poll (&tty->link->write_wait,POLLOUT ); +if (waitqueue_active (&tty->link->write_wait)) +wake_up_interruptible_poll (&tty->link->write_wait,POLLOUT );return ; }
@@ -350,7 +351,8 @@ static void n_tty_packet_mode_flush(struct tty_struct *tty)spin_lock_irqsave (&tty->ctrl_lock, flags);if (tty->link->packet) { tty->ctrl_status |=TIOCPKT_FLUSHREAD ; -wake_up_interruptible (&tty->link->read_wait); +if (waitqueue_active (&tty->link->read_wait)) +wake_up_interruptible (&tty->link->read_wait); }spin_unlock_irqrestore (&tty->ctrl_lock, flags); }
@@ -1156,7 +1158,8 @@ static void n_tty_receive_break(struct tty_struct *tty)put_tty_queue ('\0' , ldata); }put_tty_queue ('\0' , ldata); -wake_up_interruptible (&tty->read_wait); +if (waitqueue_active (&tty->read_wait)) +wake_up_interruptible (&tty->read_wait); }/**
@@ -1214,7 +1217,8 @@ static void n_tty_receive_parity_error(struct tty_struct *tty, unsigned char c)put_tty_queue ('\0' , ldata); elseput_tty_queue (c ,ldata ); -wake_up_interruptible (&tty->read_wait); +if (waitqueue_active (&tty->read_wait)) +wake_up_interruptible (&tty->read_wait); } static void
@@ -1808,8 +1812,10 @@ static void n_tty_set_termios(struct tty_struct *tty, struct ktermios *old)start_tty (tty);/* The termios change make the tty ready for I/O */ -wake_up_interruptible (&tty->write_wait); -wake_up_interruptible (&tty->read_wait); +if (waitqueue_active (&tty->write_wait)) +wake_up_interruptible (&tty->write_wait); +if (waitqueue_active (&tty->read_wait)) +wake_up_interruptible (&tty->read_wait); }/**
Signature · 1 line
1.8.1.2
Re: [PATCH tty-next 6/7] n_tty: Only perform wakeups for waiters
Quoted 12 lines · On Fri, 22 Nov 2013 10:59:24 -0500 Peter Hurley <[email protected]>…
Only wakeup the _waiting_ reader, polls and/or writer(s).
Signed-off-by Peter Hurley <[email protected]>--- drivers/tty/n_tty.c | 18 ++++++++++++------ 1 file changed, 12 insertions(+), 6 deletions(-)drivers/tty/n_tty.c +0
-0
drivers/tty/n_tty.c static void n_tty_check_unthrottle(struct tty_struct *tty) · line 279
Quoted 6 lines
drivers/tty/n_tty.c +2
-1@@ -275,7 +275,8 @@ static void n_tty_check_unthrottle(struct tty_struct *tty)return ;n_tty_set_room (tty);n_tty_write_wakeup (tty->link); -wake_up_interruptible_poll (&tty->link->write_wait,POLLOUT ); +if (waitqueue_active (&tty->link->write_wait)) +wake_up_interruptible_poll (&tty->link->write_wait,POLLOUT );
Does this actually microbenchmark faster ?
Alan
Re: [PATCH tty-next 6/7] n_tty: Only perform wakeups for waiters
Quoted 10 lines · One Thousand Gnomes
Quoted 7 lines · On Fri, 22 Nov 2013 10:59:24 -0500 Peter Hurley <[email protected]>…
Only wakeup the _waiting_ reader, polls and/or writer(s).
Signed-off-by Peter Hurley <[email protected]>--- drivers/tty/n_tty.c | 18 ++++++++++++------ 1 file changed, 12 insertions(+), 6 deletions(-)
Quoted 13 lines
drivers/tty/n_tty.c static void n_tty_check_unthrottle(struct tty_struct *tty) · line 279
Quoted 11 lines
drivers/tty/n_tty.c +2
-1@@ -275,7 +275,8 @@ static void n_tty_check_unthrottle(struct tty_struct *tty)return ;n_tty_set_room (tty);n_tty_write_wakeup (tty->link); -wake_up_interruptible_poll (&tty->link->write_wait,POLLOUT ); +if (waitqueue_active (&tty->link->write_wait)) +wake_up_interruptible_poll (&tty->link->write_wait,POLLOUT );Does this actually microbenchmark faster ?
Getting on and off the write_wait queue is actually pretty expensive for the "other" pty (the writer), and the unnecessary wakeup from the reader doesn't help.
The other chunks are gratuitous.
Regards,
Peter Hurley
PS - This came up because there is some worst-case behavior that I'm looking into fixing. When the userspace reader is very far behind (say because it's reading char-by-char), it doesn't make sense to keep restarting the input processing worker.
PATCH 7/7 n_tty: trace input/read flow control
In reply to Peter HurleyInstrument .receive_buf() and read() paths with trace_printk's to aid in debugging flow control changes.
--- drivers/tty/n_tty.c | 76 ++++++++++++++++++++++++++++++++++++++++++----------- 1 file changed, 60 insertions(+), 16 deletions(-)
drivers/tty/n_tty.c +60 -16
@@ -83,9 +83,14 @@#undef N_TTY_TRACE#ifdef N_TTY_TRACE -# define n_tty_trace(f, args...) trace_printk(f, ##args) +# define n_tty_trace(tty, f, args...) \ + do { \ + char buf[64]; \ + trace_printk("[%s] " f, tty_name(tty, buf), ##args); \ + } while (0)#else -# define n_tty_trace(f, args...) +# define n_tty_trace(tty, f, args...) \ + do { } while (0)#endif struct n_tty_data {
@@ -179,8 +184,13 @@ static int receive_room(struct tty_struct *tty)* that erase characters will be handled. Other excess * characters will be beeped. */ -if (left <=0 ) +if (left <=0 ) { left = ldata->icanon && ldata->canon_head == ldata->read_tail; +if (left) { +n_tty_trace (tty,"overflow, head:%zu tail:%zu\n" , + ldata->read_head, ldata->read_tail); + } + }return left; }
@@ -200,11 +210,14 @@ static int receive_room(struct tty_struct *tty)static void n_tty_set_room (struct tty_struct *tty) {struct n_tty_data *ldata = tty->disc_data; +int avail;/* Did this open up the receive buffer? We may need to flip */ -if (unlikely (ldata->no_room) &&receive_room (tty)) { +if (unlikely (ldata->no_room) && (avail =receive_room (tty))) { ldata->no_room =0 ; +n_tty_trace (tty,"restart worker (avail = %d)\n" , avail); +WARN_RATELIMIT (tty->port->itty ==NULL ,"scheduling with invalid itty\n" );/* see if ldisc has been killed - if so, this means that
@@ -240,6 +253,8 @@ static ssize_t chars_in_buffer(struct tty_struct *tty)static void n_tty_write_wakeup (struct tty_struct *tty) { +n_tty_trace (tty,"\n" ); +if (tty->fasync &&test_and_clear_bit (TTY_DO_WRITE_WAKEUP , &tty->flags))kill_fasync (&tty->fasync,SIGIO ,POLL_OUT ); }
@@ -254,13 +269,16 @@ static void n_tty_check_throttle(struct tty_struct *tty)* canonical mode and don't have a newline yet! */ while (1 ) { -int throttled; +int throttled, avail;tty_set_flow_change (tty,TTY_THROTTLE_SAFE ); -if (receive_room (tty) >=TTY_THRESHOLD_THROTTLE ) + avail =receive_room (tty); +if (avail >=TTY_THRESHOLD_THROTTLE )break ; throttled =tty_throttle_safe (tty); -if (!throttled) +if (!throttled) { +n_tty_trace (tty,"throttled (avail = %d)/n" , avail);break ; + } }__tty_set_flow_change (tty,0 ); }
@@ -289,16 +307,19 @@ static void n_tty_check_unthrottle(struct tty_struct *tty)*/ while (1 ) { -int unthrottled; +int unthrottled, used;tty_set_flow_change (tty,TTY_UNTHROTTLE_SAFE ); -if (chars_in_buffer (tty) >TTY_THRESHOLD_UNTHROTTLE ) + used =chars_in_buffer (tty); +if (used >TTY_THRESHOLD_UNTHROTTLE )break ;if (!tty->count)break ;n_tty_set_room (tty); unthrottled =tty_unthrottle_safe (tty); -if (!unthrottled) +if (!unthrottled) { +n_tty_trace (tty,"unthrottled (used = %d)\n" , used);break ; + } }__tty_set_flow_change (tty,0 ); }
@@ -374,6 +395,9 @@ static void n_tty_packet_mode_flush(struct tty_struct *tty)static void n_tty_flush_buffer (struct tty_struct *tty) {down_write (&tty->termios_rwsem); + +n_tty_trace (tty,"\n" ); +reset_buffer_flags (tty->disc_data);n_tty_set_room (tty);
@@ -770,6 +794,10 @@ static size_t __process_echoes(struct tty_struct *tty)* of echo overrun before the next commit), then discard enough * data at the tail to prevent a subsequent overrun */ while (ldata->echo_commit - tail >=ECHO_DISCARD_WATERMARK ) { + +n_tty_trace (tty,"discard echoes (%zu - %zu > %d)\n" , + ldata->echo_commit, tail,ECHO_DISCARD_WATERMARK ); +if (echo_buf (ldata, tail) ==ECHO_OP_START ) {if (echo_buf (ldata, tail +1 ) ==ECHO_OP_ERASE_TAB ) tail +=3 ;
@@ -1367,6 +1395,7 @@ handle_newline:set_bit (ldata->read_head & (N_TTY_BUF_SIZE -1 ), ldata->read_flags);put_tty_queue (c, ldata); ldata->canon_head = ldata->read_head; +n_tty_trace (tty,"canon_head:%zu\n" , ldata->canon_head);kill_fasync (&tty->fasync,SIGIO ,POLL_IN );if (waitqueue_active (&tty->read_wait))wake_up_interruptible (&tty->read_wait);
@@ -1668,6 +1697,7 @@ static void __receive_buf(struct tty_struct *tty, const unsigned char *cp,if ((!ldata->icanon && (read_cnt (ldata) >= ldata->minimum_to_wake)) ||L_EXTPROC (tty)) { +n_tty_trace (tty,"reader wakeup, count:%zd\n" ,read_cnt (ldata));kill_fasync (&tty->fasync,SIGIO ,POLL_IN );if (waitqueue_active (&tty->read_wait))wake_up_interruptible (&tty->read_wait);
@@ -1687,6 +1717,8 @@ n_tty_receive_buf_common(struct tty_struct *tty, const unsigned char *cp, room =receive_room (tty); n =min (count, room);if (!n) { +if (!room) +n_tty_trace (tty,"buffer full\n" );if (flow && !room) ldata->no_room =1 ;break ;
@@ -1703,6 +1735,8 @@ n_tty_receive_buf_common(struct tty_struct *tty, const unsigned char *cp,n_tty_check_throttle (tty);up_read (&tty->termios_rwsem); +n_tty_trace (tty,"rcvd:%d\n" , rcvd); +return rcvd; }
@@ -1811,6 +1845,8 @@ static void n_tty_set_termios(struct tty_struct *tty, struct ktermios *old)if (!I_IXON (tty) && old && (old->c_iflag &IXON ) && !tty->flow_stopped)start_tty (tty); +n_tty_trace (tty,"canon:%d\n" , ldata->icanon); +/* The termios change make the tty ready for I/O */ if (waitqueue_active (&tty->write_wait))wake_up_interruptible (&tty->write_wait);
@@ -1890,10 +1926,14 @@ static inline int input_available_p(struct tty_struct *tty, int poll)int amt = poll && !TIME_CHAR (tty) ?MIN_CHAR (tty) :1 ;if (ldata->icanon && !L_EXTPROC (tty)) { -if (ldata->canon_head != ldata->read_tail) +if (ldata->canon_head != ldata->read_tail) { +n_tty_trace (tty,"line avail\n" );return 1 ; - }else if (read_cnt (ldata) >= amt) + } + }else if (read_cnt (ldata) >= amt) { +n_tty_trace (tty,"avail:%zu amt:%d\n" ,read_cnt (ldata), amt);return 1 ; + }return 0 ; }
@@ -1984,8 +2024,8 @@ static int canon_copy_from_read_buf(struct tty_struct *tty, tail = ldata->read_tail & (N_TTY_BUF_SIZE -1 ); size =min_t (size_t, tail + n,N_TTY_BUF_SIZE ); -n_tty_trace ("%s: nr:%zu tail:%zu n:%zu size:%zu\n" , - __func__, *nr, tail, n, size); +n_tty_trace (tty,"nr:%zu tail:%zu n:%zu size:%zu\n" , + *nr, tail, n, size); eol =find_next_bit (ldata->read_flags, size, tail); more = n - (size - tail);
@@ -2009,8 +2049,8 @@ static int canon_copy_from_read_buf(struct tty_struct *tty, eof_push = !n && ldata->read_tail != ldata->line_start; } -n_tty_trace ("%s: eol:%zu found:%d n:%zu c:%zu size:%zu more:%zu\n" , - __func__, eol, found, n, c, size, more); +n_tty_trace (tty,"eol:%zu found:%d n:%zu c:%zu size:%zu more:%zu\n" , + eol, found, n, c, size, more);if (n > size) { ret =copy_to_user (*b,read_buf_addr (ldata, tail), size);
@@ -2202,6 +2242,8 @@ static ssize_t n_tty_read(struct tty_struct *tty, struct file *file,n_tty_set_room (tty);up_read (&tty->termios_rwsem); +n_tty_trace (tty,"blocking\n" ); + timeout =schedule_timeout (timeout);down_read (&tty->termios_rwsem);
@@ -2259,6 +2301,8 @@ static ssize_t n_tty_read(struct tty_struct *tty, struct file *file,if (b - buf) retval = b - buf; +n_tty_trace (tty,"ret:%zd\n" , retval); +return retval; }
Signature · 1 line
1.8.1.2
Re: [PATCH tty-next 7/7] n_tty: trace input/read flow control
Quoted 2 lines · On Fri, 22 Nov 2013 10:59:25 -0500 Peter Hurley <[email protected]>…
Instrument .receive_buf() and
read()paths with trace_printk's to aid in debugging flow control changes.
tty devices have a device, we have a dev_dbg() layer. The old tty trace predates this but there isn't really any excuse for not using it now that I can see ?
Alan
Re: [PATCH tty-next 7/7] n_tty: trace input/read flow control
Quoted 9 lines · One Thousand Gnomes
Quoted 2 lines · On Fri, 22 Nov 2013 10:59:25 -0500 Peter Hurley <[email protected]>…
Instrument .receive_buf() and
read()paths with trace_printk's to aid in debugging flow control changes.tty devices have a device, we have a
dev_dbg()layer. The old tty trace predates this but there isn't really any excuse for not using it now that I can see ?
I was using the ftrace facility because it has significantly less performance impact than printk (which was an important factor while debugging flow control problems).
That said, I could further macro-ize n_tty_trace() with selectable facility (which would be useful when debugging problems that end in cpu death).
Regards,
Peter Hurley
Re: [PATCH tty-next 7/7] n_tty: trace input/read flow control
Quoted 11 lines · Peter Hurley
Quoted 9 lines · One Thousand Gnomes
Quoted 2 lines · On Fri, 22 Nov 2013 10:59:25 -0500 Peter Hurley <[email protected]>…
Instrument .receive_buf() and
read()paths with trace_printk's to aid in debugging flow control changes.tty devices have a device, we have a
dev_dbg()layer. The old tty trace predates this but there isn't really any excuse for not using it now that I can see ?
Quoted 6 lines
I was using the ftrace facility because it has significantly less performance impact than printk (which was an important factor while debugging flow control problems).
That said, I could further macro-ize
n_tty_trace()with selectable facility (which would be useful when debugging problems that end in cpu death).
Greg,
I don't really mind if you don't want to take this patch; it is useful but it does clutter up the code.
Regards,
Peter Hurley
v2 cover n_tty cleanup + trace additions
In reply to Peter HurleyGreg,
This patchset is a v2 respin which contains no bug fixes, only misc cleanup for the N_TTY line discipline and significant additions to debug tracing.
As I mentioned in a previous email, I can hold on to patches 7 and 8 and keep them out-of-tree if you'd rather not have instrumented code in the tty layer.
v2:
- Patches 1-6 are unchanged from v1.
- Patch 7 changes: make n_tty_trace() configurable to output via pr_xxxx() or dev_dbg() rather than trace_printk() per Alan's comments
- Patch 8 adds echo buffer indices tracing (separately selectable)
v1:
Patches 1-4 cleans up code duplication remnants from 3.12's changes. Patch 5 simplifies the snarled logic of how much data is considered 'input available'. I was getting tired of hand checking that logic every time I touched minimum_to_wake.
Patch 6 reduces unnecessary wake ups.
Patch 7 extends the trace usage to aid in identifying and fixing input flow-control bugs (this was used to fix the readline() bug). I also expect to use it to reduce input processing restarts. It's use is disabled by default and only enabled with the file-local define N_TTY_TRACE (similar to TTY_DEBUG_HANGUP in tty_io.c).
Regards,
Peter Hurley (8):
n_tty: Merge .receive_buf() flavors
n_tty: Un-inline slow-path n_tty_receive_char()
n_tty: Un-inline slow-path n_tty_receive_char_closing()
n_tty: Refactor PARMRK doubling checks
n_tty: Refactor input_available_p() by call site
n_tty: Only perform wakeups for waiters
n_tty: Extend debug tracing
n_tty: Trace echo output with separate trace configuration
drivers/tty/n_tty.c | 174 +++++++++++++++++++++++++++++++++------------------- 1 file changed, 112 insertions(+), 62 deletions(-)
Signature · 1 line
1.8.1.2
PATCH v2 1/8 n_tty: Merge .receive_buf() flavors
N_TTY's direct and flow-controlled flavors of the .receive_buf() method are nearly identical; fold together.
--- drivers/tty/n_tty.c | 43 ++++++++++++++++--------------------------- 1 file changed, 16 insertions(+), 27 deletions(-)
drivers/tty/n_tty.c +16 -27
@@ -1677,32 +1677,9 @@ static void __receive_buf(struct tty_struct *tty, const unsigned char *cp, } } -static void n_tty_receive_buf (struct tty_struct *tty,const unsigned char *cp, -char *fp,int count) -{ -int room, n; - -down_read (&tty->termios_rwsem); - -while (1 ) { - room =receive_room (tty); - n =min (count, room); -if (!n) -break ; -__receive_buf (tty, cp, fp, n); - cp += n; -if (fp) - fp += n; - count -= n; - } - - tty->receive_room = room; -n_tty_check_throttle (tty); -up_read (&tty->termios_rwsem); -} - -static int n_tty_receive_buf2 (struct tty_struct *tty,const unsigned char *cp, -char *fp,int count) +static int +n_tty_receive_buf_common (struct tty_struct *tty,const unsigned char *cp, +char *fp,int count,int flow) {struct n_tty_data *ldata = tty->disc_data;int room, n, rcvd =0 ;
@@ -1713,7 +1690,7 @@ static int n_tty_receive_buf2(struct tty_struct *tty, const unsigned char *cp, room =receive_room (tty); n =min (count, room);if (!n) { -if (!room) +if (flow && !room) ldata->no_room =1 ;break ; }
@@ -1732,6 +1709,18 @@ static int n_tty_receive_buf2(struct tty_struct *tty, const unsigned char *cp,return rcvd; } +static void n_tty_receive_buf (struct tty_struct *tty,const unsigned char *cp, +char *fp,int count) +{ +n_tty_receive_buf_common (tty, cp, fp, count,0 ); +} + +static int n_tty_receive_buf2 (struct tty_struct *tty,const unsigned char *cp, +char *fp,int count) +{ +return n_tty_receive_buf_common (tty, cp, fp, count,1 ); +} +int is_ignored (int sig) { return (sigismember(¤t->blocked, sig) ||
Signature · 1 line
1.8.1.2
PATCH v2 2/8 n_tty: Un-inline slow-path n_tty_receive_char()
In reply to Peter HurleyCommit e60d27c4d8b33ba20896b76b6558f061bc6460ff,n_tty: Factor LNEXT processing from per-char i/o path, mistakenly inlined the non-inline alias, n_tty_receive_char(), for the inline function, n_tty_receive_char_inline().
As n_tty_receive_char() is intended for slow-path char processing only, un-inline it.
--- drivers/tty/n_tty.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/tty/n_tty.c +1 -1
@@ -1420,7 +1420,7 @@ n_tty_receive_char_inline(struct tty_struct *tty, unsigned char c)put_tty_queue (c, ldata); } -static inline void n_tty_receive_char (struct tty_struct *tty,unsigned char c) +static void n_tty_receive_char (struct tty_struct *tty,unsigned char c) {n_tty_receive_char_inline (tty, c); }
Signature · 1 line
1.8.1.2
PATCH v2 3/8 n_tty: Un-inline slow-path n_tty_receive_char_closing()
In reply to Peter HurleyAlthough n_tty_receive_char_closing() only has one call-site, let the compiler inline instead.
--- drivers/tty/n_tty.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/tty/n_tty.c +1 -2
@@ -1445,8 +1445,7 @@ n_tty_receive_char_fast(struct tty_struct *tty, unsigned char c)put_tty_queue (c, ldata); } -static inline void -n_tty_receive_char_closing (struct tty_struct *tty,unsigned char c) +static void n_tty_receive_char_closing (struct tty_struct *tty,unsigned char c) {if (I_ISTRIP (tty)) c &=0x7f ;
Signature · 1 line
1.8.1.2
PATCH v2 4/8 n_tty: Refactor PARMRK doubling checks
In reply to Peter HurleyPerform PARMRK doubling checks explicitly; remove ternary idiom and local variable.
--- drivers/tty/n_tty.c | 14 +++++--------- 1 file changed, 5 insertions(+), 9 deletions(-)
drivers/tty/n_tty.c +5 -9
@@ -1260,7 +1260,6 @@ static int n_tty_receive_char_special(struct tty_struct *tty, unsigned char c) {struct n_tty_data *ldata = tty->disc_data; -int parmrk;if (I_IXON (tty)) {if (c ==START_CHAR (tty)) {
@@ -1345,8 +1344,6 @@ n_tty_receive_char_special(struct tty_struct *tty, unsigned char c) }if ((c ==EOL_CHAR (tty)) || (c ==EOL2_CHAR (tty) &&L_IEXTEN (tty))) { - parmrk = (c == (unsigned char )'\377' &&I_PARMRK (tty)) - ?1 :0 ;/* * XXX are EOL_CHAR and EOL2_CHAR echoed?!? */
@@ -1361,7 +1358,7 @@ n_tty_receive_char_special(struct tty_struct *tty, unsigned char c)* XXX does PARMRK doubling happen for * EOL_CHAR and EOL2_CHAR? */ -if (parmrk) +if (c == (unsigned char )'\377' &&I_PARMRK (tty))put_tty_queue (c, ldata); handle_newline:
@@ -1375,7 +1372,6 @@ handle_newline: } } - parmrk = (c == (unsigned char )'\377' &&I_PARMRK (tty)) ?1 :0 ;if (L_ECHO (tty)) {finish_erasing (ldata);if (c =='\n' )
@@ -1389,7 +1385,8 @@ handle_newline:commit_echoes (tty); } -if (parmrk) +/* PARMRK doubling check */ +if (c == (unsigned char )'\377' &&I_PARMRK (tty))put_tty_queue (c, ldata);put_tty_queue (c, ldata);
@@ -1400,7 +1397,6 @@ static inline void n_tty_receive_char_inline(struct tty_struct *tty, unsigned char c) {struct n_tty_data *ldata = tty->disc_data; -int parmrk;if (tty->stopped && !tty->flow_stopped &&I_IXON (tty) &&I_IXANY (tty)) {start_tty (tty);
@@ -1414,8 +1410,8 @@ n_tty_receive_char_inline(struct tty_struct *tty, unsigned char c)echo_char (c, tty);commit_echoes (tty); } - parmrk = (c == (unsigned char )'\377' &&I_PARMRK (tty)) ?1 :0 ; -if (parmrk) +/* PARMRK doubling check */ +if (c == (unsigned char )'\377' &&I_PARMRK (tty))put_tty_queue (c, ldata);put_tty_queue (c, ldata); }
Signature · 1 line
1.8.1.2
PATCH v2 5/8 n_tty: Refactor input_available_p() by call site
In reply to Peter HurleyDistinguish if caller is n_tty_poll() or n_tty_read(), and set the read/wakeup threshold accordingly.
--- drivers/tty/n_tty.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-)
drivers/tty/n_tty.c +4 -3
@@ -1881,14 +1881,15 @@ static int n_tty_open(struct tty_struct *tty)return 0 ; } -static inline int input_available_p (struct tty_struct *tty,int amt) +static inline int input_available_p (struct tty_struct *tty,int poll) {struct n_tty_data *ldata = tty->disc_data; +int amt = poll && !TIME_CHAR (tty) ?MIN_CHAR (tty) :1 ;if (ldata->icanon && !L_EXTPROC (tty)) {if (ldata->canon_head != ldata->read_tail)return 1 ; - }else if (read_cnt (ldata) >= (amt ? amt :1 )) + }else if (read_cnt (ldata) >= amt)return 1 ;return 0 ;
@@ -2393,7 +2394,7 @@ static unsigned int n_tty_poll(struct tty_struct *tty, struct file *file,poll_wait (file, &tty->read_wait, wait);poll_wait (file, &tty->write_wait, wait); -if (input_available_p (tty,TIME_CHAR (tty) ?0 :MIN_CHAR (tty))) +if (input_available_p (tty,1 )) mask |=POLLIN |POLLRDNORM ;if (tty->packet && tty->link->ctrl_status) mask |=POLLPRI |POLLIN |POLLRDNORM ;
Signature · 1 line
1.8.1.2
PATCH v2 6/8 n_tty: Only perform wakeups for waiters
In reply to Peter HurleyOnly wakeup the _waiting_ reader, polls and/or writer(s).
--- drivers/tty/n_tty.c | 18 ++++++++++++------ 1 file changed, 12 insertions(+), 6 deletions(-)
drivers/tty/n_tty.c +12 -6
@@ -275,7 +275,8 @@ static void n_tty_check_unthrottle(struct tty_struct *tty)return ;n_tty_set_room (tty);n_tty_write_wakeup (tty->link); -wake_up_interruptible_poll (&tty->link->write_wait,POLLOUT ); +if (waitqueue_active (&tty->link->write_wait)) +wake_up_interruptible_poll (&tty->link->write_wait,POLLOUT );return ; }
@@ -350,7 +351,8 @@ static void n_tty_packet_mode_flush(struct tty_struct *tty)spin_lock_irqsave (&tty->ctrl_lock, flags);if (tty->link->packet) { tty->ctrl_status |=TIOCPKT_FLUSHREAD ; -wake_up_interruptible (&tty->link->read_wait); +if (waitqueue_active (&tty->link->read_wait)) +wake_up_interruptible (&tty->link->read_wait); }spin_unlock_irqrestore (&tty->ctrl_lock, flags); }
@@ -1158,7 +1160,8 @@ static void n_tty_receive_break(struct tty_struct *tty)put_tty_queue ('\0' , ldata); }put_tty_queue ('\0' , ldata); -wake_up_interruptible (&tty->read_wait); +if (waitqueue_active (&tty->read_wait)) +wake_up_interruptible (&tty->read_wait); }/**
@@ -1216,7 +1219,8 @@ static void n_tty_receive_parity_error(struct tty_struct *tty, unsigned char c)put_tty_queue ('\0' , ldata); elseput_tty_queue (c ,ldata ); -wake_up_interruptible (&tty->read_wait); +if (waitqueue_active (&tty->read_wait)) +wake_up_interruptible (&tty->read_wait); } static void
@@ -1811,8 +1815,10 @@ static void n_tty_set_termios(struct tty_struct *tty, struct ktermios *old)start_tty (tty);/* The termios change make the tty ready for I/O */ -wake_up_interruptible (&tty->write_wait); -wake_up_interruptible (&tty->read_wait); +if (waitqueue_active (&tty->write_wait)) +wake_up_interruptible (&tty->write_wait); +if (waitqueue_active (&tty->read_wait)) +wake_up_interruptible (&tty->read_wait); }/**
Signature · 1 line
1.8.1.2
PATCH v2 7/8 n_tty: Extend debug tracing
In reply to Peter HurleyAdd configurable trace support to allow tracing with dev_dbg, pr_dbg, or trace_printk. Instrument .receive_buf() and read() paths with traces to aid in debugging flow control changes.
--- drivers/tty/n_tty.c | 79 ++++++++++++++++++++++++++++++++++++++++++----------- 1 file changed, 63 insertions(+), 16 deletions(-)
drivers/tty/n_tty.c +63 -16
@@ -80,12 +80,18 @@#define ECHO_BLOCK 256#define ECHO_DISCARD_WATERMARK N_TTY_BUF_SIZE - (ECHO_BLOCK + 32) +#define n_tty_trace_facility trace_printk +#define _n_tty_trace(tty, f, args...) \ + do { \ + char buf[64]; \ + n_tty_trace_facility("[%s] " f, tty_name(tty, buf), ##args); \ + } while (0)#undef N_TTY_TRACE#ifdef N_TTY_TRACE -# define n_tty_trace(f, args...) trace_printk(f, ##args) +# define n_tty_trace(tty, f, args...) _n_tty_trace(tty, f, args)#else -# define n_tty_trace(f, args...) +# define n_tty_trace(tty, f, args...) do { } while (0)#endif struct n_tty_data {
@@ -179,8 +185,13 @@ static int receive_room(struct tty_struct *tty)* that erase characters will be handled. Other excess * characters will be beeped. */ -if (left <=0 ) +if (left <=0 ) { left = ldata->icanon && ldata->canon_head == ldata->read_tail; +if (left) { +n_tty_trace (tty,"overflow, head:%zu tail:%zu\n" , + ldata->read_head, ldata->read_tail); + } + }return left; }
@@ -200,11 +211,14 @@ static int receive_room(struct tty_struct *tty)static void n_tty_set_room (struct tty_struct *tty) {struct n_tty_data *ldata = tty->disc_data; +int avail;/* Did this open up the receive buffer? We may need to flip */ -if (unlikely (ldata->no_room) &&receive_room (tty)) { +if (unlikely (ldata->no_room) && (avail =receive_room (tty))) { ldata->no_room =0 ; +n_tty_trace (tty,"restart worker (avail = %d)\n" , avail); +WARN_RATELIMIT (tty->port->itty ==NULL ,"scheduling with invalid itty\n" );/* see if ldisc has been killed - if so, this means that
@@ -240,6 +254,8 @@ static ssize_t chars_in_buffer(struct tty_struct *tty)static void n_tty_write_wakeup (struct tty_struct *tty) { +n_tty_trace (tty,"\n" ); +if (tty->fasync &&test_and_clear_bit (TTY_DO_WRITE_WAKEUP , &tty->flags))kill_fasync (&tty->fasync,SIGIO ,POLL_OUT ); }
@@ -254,13 +270,16 @@ static void n_tty_check_throttle(struct tty_struct *tty)* canonical mode and don't have a newline yet! */ while (1 ) { -int throttled; +int throttled, avail;tty_set_flow_change (tty,TTY_THROTTLE_SAFE ); -if (receive_room (tty) >=TTY_THRESHOLD_THROTTLE ) + avail =receive_room (tty); +if (avail >=TTY_THRESHOLD_THROTTLE )break ; throttled =tty_throttle_safe (tty); -if (!throttled) +if (!throttled) { +n_tty_trace (tty,"throttled (avail = %d)/n" , avail);break ; + } }__tty_set_flow_change (tty,0 ); }
@@ -289,16 +308,19 @@ static void n_tty_check_unthrottle(struct tty_struct *tty)*/ while (1 ) { -int unthrottled; +int unthrottled, used;tty_set_flow_change (tty,TTY_UNTHROTTLE_SAFE ); -if (chars_in_buffer (tty) >TTY_THRESHOLD_UNTHROTTLE ) + used =chars_in_buffer (tty); +if (used >TTY_THRESHOLD_UNTHROTTLE )break ;if (!tty->count)break ;n_tty_set_room (tty); unthrottled =tty_unthrottle_safe (tty); -if (!unthrottled) +if (!unthrottled) { +n_tty_trace (tty,"unthrottled (used = %d)\n" , used);break ; + } }__tty_set_flow_change (tty,0 ); }
@@ -374,6 +396,9 @@ static void n_tty_packet_mode_flush(struct tty_struct *tty)static void n_tty_flush_buffer (struct tty_struct *tty) {down_write (&tty->termios_rwsem); + +n_tty_trace (tty,"\n" ); +reset_buffer_flags (tty->disc_data);n_tty_set_room (tty);
@@ -770,6 +795,10 @@ static size_t __process_echoes(struct tty_struct *tty)* of echo overrun before the next commit), then discard enough * data at the tail to prevent a subsequent overrun */ while (ldata->echo_commit - tail >=ECHO_DISCARD_WATERMARK ) { + +n_tty_trace (tty,"discard echoes (%zu - %zu > %d)\n" , + ldata->echo_commit, tail,ECHO_DISCARD_WATERMARK ); +if (echo_buf (ldata, tail) ==ECHO_OP_START ) {if (echo_buf (ldata, tail +1 ) ==ECHO_OP_ERASE_TAB ) tail +=3 ;
@@ -1369,6 +1398,7 @@ handle_newline:set_bit (ldata->read_head & (N_TTY_BUF_SIZE -1 ), ldata->read_flags);put_tty_queue (c, ldata); ldata->canon_head = ldata->read_head; +n_tty_trace (tty,"canon_head:%zu\n" , ldata->canon_head);kill_fasync (&tty->fasync,SIGIO ,POLL_IN );if (waitqueue_active (&tty->read_wait))wake_up_interruptible (&tty->read_wait);
@@ -1670,6 +1700,7 @@ static void __receive_buf(struct tty_struct *tty, const unsigned char *cp,if ((!ldata->icanon && (read_cnt (ldata) >= ldata->minimum_to_wake)) ||L_EXTPROC (tty)) { +n_tty_trace (tty,"reader wakeup, count:%zd\n" ,read_cnt (ldata));kill_fasync (&tty->fasync,SIGIO ,POLL_IN );if (waitqueue_active (&tty->read_wait))wake_up_interruptible (&tty->read_wait);
@@ -1689,6 +1720,8 @@ n_tty_receive_buf_common(struct tty_struct *tty, const unsigned char *cp, room =receive_room (tty); n =min (count, room);if (!n) { +if (!room) +n_tty_trace (tty,"buffer full\n" );if (flow && !room) ldata->no_room =1 ;break ;
@@ -1705,6 +1738,8 @@ n_tty_receive_buf_common(struct tty_struct *tty, const unsigned char *cp,n_tty_check_throttle (tty);up_read (&tty->termios_rwsem); +n_tty_trace (tty,"rcvd:%d\n" , rcvd); +return rcvd; }
@@ -1814,6 +1849,10 @@ static void n_tty_set_termios(struct tty_struct *tty, struct ktermios *old)if (!I_IXON (tty) && old && (old->c_iflag &IXON ) && !tty->flow_stopped)start_tty (tty); +n_tty_trace (tty,"mode:%s tail:%zu canon:%zu head:%zu\n" , + ldata->icanon ?"canon" :"raw" , ldata->read_tail, + ldata->canon_head, ldata->read_head); +/* The termios change make the tty ready for I/O */ if (waitqueue_active (&tty->write_wait))wake_up_interruptible (&tty->write_wait);
@@ -1893,10 +1932,14 @@ static inline int input_available_p(struct tty_struct *tty, int poll)int amt = poll && !TIME_CHAR (tty) ?MIN_CHAR (tty) :1 ;if (ldata->icanon && !L_EXTPROC (tty)) { -if (ldata->canon_head != ldata->read_tail) +if (ldata->canon_head != ldata->read_tail) { +n_tty_trace (tty,"line avail\n" );return 1 ; - }else if (read_cnt (ldata) >= amt) + } + }else if (read_cnt (ldata) >= amt) { +n_tty_trace (tty,"avail:%zu amt:%d\n" ,read_cnt (ldata), amt);return 1 ; + }return 0 ; }
@@ -1993,8 +2036,8 @@ static int canon_copy_from_read_buf(struct tty_struct *tty, tail = ldata->read_tail & (N_TTY_BUF_SIZE -1 ); size =min_t (size_t, tail + n,N_TTY_BUF_SIZE ); -n_tty_trace ("%s: nr:%zu tail:%zu n:%zu size:%zu\n" , - __func__, *nr, tail, n, size); +n_tty_trace (tty,"nr:%zu tail:%zu n:%zu size:%zu\n" , + *nr, tail, n, size); eol =find_next_bit (ldata->read_flags, size, tail); more = n - (size - tail);
@@ -2018,8 +2061,8 @@ static int canon_copy_from_read_buf(struct tty_struct *tty, eof_push = !n && ldata->read_tail != ldata->line_start; } -n_tty_trace ("%s: eol:%zu found:%d n:%zu c:%zu size:%zu more:%zu\n" , - __func__, eol, found, n, c, size, more); +n_tty_trace (tty,"eol:%zu found:%d n:%zu c:%zu size:%zu more:%zu\n" , + eol, found, n, c, size, more);if (n > size) { ret =copy_to_user (*b,read_buf_addr (ldata, tail), size);
@@ -2211,6 +2254,8 @@ static ssize_t n_tty_read(struct tty_struct *tty, struct file *file,n_tty_set_room (tty);up_read (&tty->termios_rwsem); +n_tty_trace (tty,"blocking\n" ); + timeout =schedule_timeout (timeout);down_read (&tty->termios_rwsem);
@@ -2268,6 +2313,8 @@ static ssize_t n_tty_read(struct tty_struct *tty, struct file *file,if (b - buf) retval = b - buf; +n_tty_trace (tty,"ret:%zd\n" , retval); +return retval; }
Signature · 1 line
1.8.1.2
Greg Kroah-Hartman
maintainerRe: [PATCH v2 tty-next 7/8] n_tty: Extend debug tracing
Quoted 8 lines · Peter Hurley
Add configurable trace support to allow tracing with
dev_dbg,pr_dbg, ortrace_printk. Instrument .receive_buf() andread()paths with traces to aid in debugging flow control changes.Signed-off-by Peter Hurley <[email protected]>--- drivers/tty/n_tty.c | 79 ++++++++++++++++++++++++++++++++++++++++++----------- 1 file changed, 63 insertions(+), 16 deletions(-)
If we are going to do tracing in the tty layer, why not use the in-kernel tracing functionality instead of having this mis-mash of different ways to get tracing data out?
I'd much rather prefer that, so I've stopped applying things in this series at this point, I've applied the 6 previous ones.
thanks,
greg k-h
PATCH v2 8/8 n_tty: Trace echo output with separate trace configuration
In reply to Peter HurleyAdd trace support for echo output but leave it separately selectable/ buildable from the base trace support.
--- drivers/tty/n_tty.c | 16 ++++++++++++++-- 1 file changed, 14 insertions(+), 2 deletions(-)
drivers/tty/n_tty.c +14 -2
@@ -94,6 +94,13 @@ # define n_tty_trace(tty, f, args...) do { } while (0)#endif +#undef N_TTY_TRACE_ECHOES +#ifdef N_TTY_TRACE_ECHOES +# define n_tty_trace_echoes(tty, f, args...) _n_tty_trace(tty, f, args) +#else +# define n_tty_trace_echoes(tty, f, args...) do { } while (0) +#endif +struct n_tty_data {/* producer-published */ size_t read_head;
@@ -791,13 +798,18 @@ static size_t __process_echoes(struct tty_struct *tty) } } +n_tty_trace_echoes (tty,"echoes:%zu,%zu commit:%zu head:%zu\n" , + ldata->echo_tail, tail, ldata->echo_commit, + ldata->echo_head); +/* If the echo buffer is nearly full (so that the possibility exists * of echo overrun before the next commit), then discard enough * data at the tail to prevent a subsequent overrun */ while (ldata->echo_commit - tail >=ECHO_DISCARD_WATERMARK ) { -n_tty_trace (tty,"discard echoes (%zu - %zu > %d)\n" , - ldata->echo_commit, tail,ECHO_DISCARD_WATERMARK ); +n_tty_trace_echoes (tty,"discard echoes (%zu - %zu > %d)\n" , + ldata->echo_commit, tail, +ECHO_DISCARD_WATERMARK );if (echo_buf (ldata, tail) ==ECHO_OP_START ) {if (echo_buf (ldata, tail +1 ) ==ECHO_OP_ERASE_TAB )
Signature · 1 line
1.8.1.2