Hi Filipe, On 07/07/2015 05:28 PM, Filipe Brandenburger wrote: > On Wed, Jun 10, 2015 at 10:11 AM, Peter Hurley <peter@xxxxxxxxxxxxxxxxxx> wrote: >> Well, I still couldn't repro this bug but I have a hypothesis that >> the bash sig handler is executing before the flush occurs, perhaps >> because the flush is stalled waiting for the write lock. >> >> Would you please try the patch below? > > I just wanted to get back to you to say I've been testing your patch > for a few weeks and with the patch applied I have never reproduced the > problem with Ctrl+C truncating the output (while I've been able to > consistently reproduce it without the patch.) > > I tested it on top of 4.0.4, 4.0.5, 4.1-rc6, 4.1-rc8, 4.1, 4.1.y and 4.2-rc1. > > Do you think you could send this patch for inclusion in kernel 4.2? I missed the initial merge window for 4.2, but Greg already has this patch so will be included in linux-next, probably by the weekend. [I see now that the patch I sent to Greg wasn't cc'd to you. Sorry about that; I overlooked that the Reported-by tag didn't add you to the send list. My bad. I'll forward you Greg's upstream notification when I get it.] > Feel free to include: > > Reported-by: Filipe Brandenburger <filbranden@xxxxxxxxxx> > Tested-by: Filipe Brandenburger <filbranden@xxxxxxxxxx> > Acked-by: Filipe Brandenburger <filbranden@xxxxxxxxxx> > (whichever ones you find more appropriate) Already done. > Ideally also a reference to the Bugzilla: > https://bugzilla.kernel.org/show_bug.cgi?id=99351 Yep, already done. > And possibly one to this mail thread: > http://www.spinics.net/lists/linux-serial/msg17791.html Typically, things like this aren't added unless there was a lot of time in between the problem and the solution (months). > BTW, perhaps this also belongs in stable kernels 4.0.y and 4.1.y, if > you feel that's appropriate perhaps forward it to stable when you push > it upstream. I added a Fixes tag so this patch will make it's way to those kernels eventually (probably during this rc cycle). At that time, this patch will likely also be added to 4.2-rcX as well. Regards, Peter Hurley > > Cheers! > Filipe > > >> --- >% --- >> Subject: [PATCH] n_tty: signal and flush atomically >> >> When handling signalling char, claim the termios write lock before >> signalling waiting readers and writers to prevent further i/o >> before flushing the echo and output buffers. This prevents a >> userspace signal handler which may output from racing the terminal >> flush. >> >> Reference: Bugzilla #99351 ("Output truncated in ssh session after...") >> Fixes: commit d2b6f44779d3 ("n_tty: Fix signal handling flushes") >> Reported-by: Filipe Brandenburger <filbranden@xxxxxxxxxx> >> Signed-off-by: Peter Hurley <peter@xxxxxxxxxxxxxxxxxx> >> --- >> drivers/tty/n_tty.c | 16 +++++++++++++--- >> 1 file changed, 13 insertions(+), 3 deletions(-) >> >> diff --git a/drivers/tty/n_tty.c b/drivers/tty/n_tty.c >> index c9c27f6..ee8bfac 100644 >> --- a/drivers/tty/n_tty.c >> +++ b/drivers/tty/n_tty.c >> @@ -1108,19 +1108,29 @@ static void eraser(unsigned char c, struct tty_struct *tty) >> * Locking: ctrl_lock >> */ >> >> -static void isig(int sig, struct tty_struct *tty) >> +static void __isig(int sig, struct tty_struct *tty) >> { >> - struct n_tty_data *ldata = tty->disc_data; >> struct pid *tty_pgrp = tty_get_pgrp(tty); >> if (tty_pgrp) { >> kill_pgrp(tty_pgrp, sig, 1); >> put_pid(tty_pgrp); >> } >> +} >> >> - if (!L_NOFLSH(tty)) { >> +static void isig(int sig, struct tty_struct *tty) >> +{ >> + struct n_tty_data *ldata = tty->disc_data; >> + >> + if (L_NOFLSH(tty)) { >> + /* signal only */ >> + __isig(sig, tty); >> + >> + } else { /* signal and flush */ >> up_read(&tty->termios_rwsem); >> down_write(&tty->termios_rwsem); >> >> + __isig(sig, tty); >> + >> /* clear echo buffer */ >> mutex_lock(&ldata->output_lock); >> ldata->echo_head = ldata->echo_tail = 0; >> -- >> 2.4.3 -- To unsubscribe from this list: send the line "unsubscribe linux-serial" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
