more usb sysrq improvements
Subsystems (3)
cover more usb sysrq improvements
This patch series is to add sysrq support for the pl2303 usb device and to stop sysrq processing on non-console usb serial ports.
Thanks,
Jason.
The short log and git info is below, which is against the 2.6.30 development tree with the previous 4 accepted usb patches in the gregkh queue.
The following changes since commit a425a638c858fd10370b573bde81df3ba500e271:
are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/jwessel/linux-2.6-kgdb.git for_gregkh_2
Jason Wessel (2):
pl2303 usb_serial: implement sysrq handling on break
usb_serial: only allow sysrq on a console port
drivers/usb/serial/generic.c | 2 +- drivers/usb/serial/pl2303.c | 5 ++++- 2 files changed, 5 insertions(+), 2 deletions(-)
PATCH 1/2 pl2303 usb_serial: implement sysrq handling on break
Add callbacks to process the sysrq when using a pl2303 usb device as a console.
--- drivers/usb/serial/pl2303.c | 5 ++++- 1 files changed, 4 insertions(+), 1 deletions(-)
drivers/usb/serial/pl2303.c +4 -1
@@ -946,6 +946,8 @@ static void pl2303_update_line_status(struct usb_serial_port *port,spin_lock_irqsave (&priv->lock, flags); priv->line_status = data[status_idx];spin_unlock_irqrestore (&priv->lock, flags); +if (priv->line_status &UART_BREAK_ERROR ) +usb_serial_handle_break (port);wake_up_interruptible (&priv->delta_msr_wait); }
@@ -1056,7 +1058,8 @@ static void pl2303_read_bulk_callback(struct urb *urb)if (line_status &UART_OVERRUN_ERROR )tty_insert_flip_char (tty,0 ,TTY_OVERRUN );for (i =0 ; i < urb->actual_length; ++i) -tty_insert_flip_char (tty, data[i], tty_flag); +if (!usb_serial_handle_sysrq_char (port, data[i])) +tty_insert_flip_char (tty, data[i], tty_flag);tty_flip_buffer_push (tty); }tty_kref_put (tty);
Signature · 1 line
1.6.3.1.9.g95405b
PATCH 2/2 usb_serial: only allow sysrq on a console port
The only time a sysrq should get processed is if the attached device is a console. This is intended to protect sysrq execution on a host connected with a terminal program.
Here is the problem scenario:
host A <-- rs232 link --> host B
Host A is using mincom and a usb pl2303 device to connect to host b which is a linux system with a usb pl2303 device acting as the serial console. When host B is rebooted the pl2303 emits random junk characters on reset. These character sequences contain serial break signals most of the time and when translated to a sysrq have caused host A to get random processes killed, reboots or power down.
It is true that in this setup with this patch host B might still have the same problem as host A if you reboot host A. In most cases host A is a development host which seldom gets rebooted, and you could turn off sysrq temporarily on host B if you need to reboot host A.
--- drivers/usb/serial/generic.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
drivers/usb/serial/generic.c +1 -1
@@ -529,7 +529,7 @@ void usb_serial_generic_unthrottle(struct tty_struct *tty)int usb_serial_handle_sysrq_char (struct usb_serial_port *port,unsigned int ch) { -if (port->sysrq) { +if (port->sysrq && port->console) {if (ch &&time_before (jiffies, port->sysrq)) {handle_sysrq (ch,tty_port_tty_get (&port->port)); port->sysrq =0 ;
Signature · 1 line
1.6.3.1.9.g95405b
Re: [PATCH 2/2] usb_serial: only allow sysrq on a console port
Quoted 3 lines · On Fri, 29 May 2009 13:34:17 -0500 Jason Wessel <[email protected]>…
The only time a sysrq should get processed is if the attached device is a console. This is intended to protect sysrq execution on a host connected with a terminal program.
This doesn't seem to match any tree I can find ?
Re: [PATCH 2/2] usb_serial: only allow sysrq on a console port
On Mon 1 Jun 2009 09:35, Alan Cox pondered:
Quoted 8 lines
Quoted 3 lines · On Fri, 29 May 2009 13:34:17 -0500 Jason Wessel <[email protected]>…
The only time a sysrq should get processed is if the attached device is a console. This is intended to protect sysrq execution on a host connected with a terminal program.
This doesn't seem to match any tree I can find ?
Alan:
If Jason's patch is necessary () - should this be fixed up for standard UARTs too?
Make sure that only serial console (not _any_ serial port) responds to sysrq (or should something else be ensuring that this isn't set when the port !console? (I didn't see anything in serial_core.c?)
---
include/linux/serial_core.h +1 -1
@@ -474,7 +474,7 @@ static inline int uart_handle_sysrq_char(struct uart_port *port, unsigned int ch) {#ifdef SUPPORT_SYSRQ -if (port->sysrq) { +if (port->sysrq && port->console) {if (ch &&time_before (jiffies, port->sysrq)) {handle_sysrq (ch, port->info->port.tty); port->sysrq =0 ;
----------
That brings up the next question...
The above patch would sync the (seemlying duplicated) code between drivers/usb/serial/generic.c and include/linux/serial_core.h
Greg?
drivers/usb/serial/generic.c
int usb_serial_handle_sysrq_char(struct usb_serial_port *port, unsigned int ch)
{
if (port->sysrq && port->console) {
if (ch && time_before(jiffies, port->sysrq)) {
handle_sysrq(ch, tty_port_tty_get(&port->port));
port->sysrq = 0;
return 1;
}
port->sysrq = 0;
}
return 0;
}include/linux/serial_core.h
static inline intuart_handle_sysrq_char(struct uart_port *port, unsigned int ch) {
#ifdef SUPPORT_SYSRQ
if (port->sysrq) {
if (ch && time_before(jiffies, port->sysrq)) {
handle_sysrq(ch, port->info->port.tty);
port->sysrq = 0;
return 1;
}
port->sysrq = 0;
}
#endif
return 0;
}Re: [PATCH 2/2] usb_serial: only allow sysrq on a console port
Quoted 2 lines
If Jason's patch is necessary () - should this be fixed up for standard UARTs too?
I think so yes, although I'd not realised it wasn't protected currently ?
Quoted 4 lines
Make sure that only serial console (not _any_ serial port) responds to sysrq (or should something else be ensuring that this isn't set when the port !console? (I didn't see anything in
serial_core.c?)
will check
Quoted 2 lines
The above patch would sync the (seemlying duplicated) code between
drivers/usb/serial/generic.candinclude/linux/serial_core.h
There is a lot of near duplicate code like this. That is one reason for adding struct tty_port. In theory both could be collapsed into
int tty_port_handle_sysrq(struct tty_port *port, unsigned int ch)
{
}at this point as both USB and serial layer UARTs now have a port object. That would just need port->sysrq collapsing into the tty_port. port->console sort of already is.
Quoted 10 lines
Greg?
drivers/usb/serial/generic.c
intusb_serial_handle_sysrq_char(struct usb_serial_port *port, unsigned int ch){ if (port->sysrq && port->console) { if (ch && time_before(jiffies, port->sysrq)) { handle_sysrq(ch, tty_port_tty_get(&port->port)); port->sysrq = 0; return 1;
That also looks wrong - tty_port_tty_get takes a tty kref. I will check that Monday and look at collapsing these int one as you note.
Alan
Re: [PATCH 2/2] usb_serial: only allow sysrq on a console port
On Sun 5 Jul 2009 14:17, Alan Cox pondered:
Quoted 5 lines
Quoted 2 lines
If Jason's patch is necessary () - should this be fixed up for standard UARTs too?
I think so yes, although I'd not realised it wasn't protected currently ?
Hmm - try as I may - I can't get this to fail - so it must be protected somewhere....
Ahh---
It is in include/linux/serial_core.h:uart_handle_break() - never checked the header for the magic before I bugged you... sorry about that...
Quoted 6 lines
Quoted 2 lines
The above patch would sync the (seemlying duplicated) code between
drivers/usb/serial/generic.candinclude/linux/serial_core.hThere is a lot of near duplicate code like this. That is one reason for adding struct
tty_port. In theory both could be collapsed into
Quoted 7 lines
int tty_port_handle_sysrq(struct tty_port *port, unsigned int ch) { }at this point as both USB and serial layer UARTs now have a port object. That would just need
port->sysrqcollapsing into thetty_port.port->consolesort of already is.
It appears that the usb serial doesn't handle breaks like serial_core does. (I don't see any support for SAK in usb_serial either?)
Maybe _that_ is the real problem that Jason is trying to work around???
Quoted 13 lines
Quoted 2 lines
drivers/usb/serial/generic.c
int usb_serial_handle_sysrq_char(structusb_serial_port*port,unsigned int ch)
Quoted 4 lines
{ if (port->sysrq && port->console) { if (ch && time_before(jiffies, port->sysrq)) { handle_sysrq(ch,
tty_port_tty_get(&port->port));Quoted 2 lines
port->sysrq = 0; return 1;That also looks wrong -
tty_port_tty_gettakes a tty kref. I will check that Monday and look at collapsing these int one as you note.
Re: [PATCH 2/2] usb_serial: only allow sysrq on a console port
Quoted 25 lines · Robin Getz
On Sun 5 Jul 2009 14:17, Alan Cox pondered:
Quoted 5 lines
Quoted 2 lines
If Jason's patch is necessary () - should this be fixed up for standard UARTs too?
I think so yes, although I'd not realised it wasn't protected currently ?
Hmm - try as I may - I can't get this to fail - so it must be protected somewhere....
Ahh---
It is in include/linux/serial_core.h:uart_handle_break() - never checked the header for the magic before I bugged you... sorry about that...
Quoted 10 lines
Quoted 2 lines
The above patch would sync the (seemlying duplicated) code between
drivers/usb/serial/generic.candinclude/linux/serial_core.hThere is a lot of near duplicate code like this. That is one reason for adding struct
tty_port. In theory both could be collapsed intoint
tty_port_handle_sysrq(struct tty_port *port, unsigned int ch){
}
Quoted 8 lines
Quoted 3 lines
at this point as both USB and serial layer UARTs now have a port object. That would just need
port->sysrqcollapsing into thetty_port.port->consolesort of already is.It appears that the usb serial doesn't handle breaks like
serial_coredoes. (I don't see any support for SAK inusb_serialeither?)Maybe _that_ is the real problem that Jason is trying to work around???
perhaps, but what Jason proposed originally sounds pretty sane. if i enable sysrq support on my desktop, i dont want some development board being able to send a sysrq request back over my serial port and rebooting my desktop.
-mike
Re: [PATCH 2/2] usb_serial: only allow sysrq on a console port
On Mon 6 Jul 2009 00:56, Mike Frysinger pondered:
Quoted 7 lines · Robin Getz
Quoted 4 lines
It appears that the usb serial doesn't handle breaks like
serial_coredoes. (I don't see any support for SAK inusb_serialeither?)Maybe _that_ is the real problem that Jason is trying to work around???
perhaps, but what Jason proposed originally sounds pretty sane.
Which is why Greg added it to the USB tree, and it is in 2.6.31-rc2 :)
Quoted 4 lines
if i
enable sysrq support on my desktop, i dont want some development board being able to send a sysrq request back over my serial port and rebooting my desktop.
Fixing something in the wrong place -- while noble -- is still wrong :)
However -- with the way that usb serial is structured - I didn't see a better place right now either.
The point being that is would be nice to make usb serial look more like other serial devices - so they can share mode code - and things like this (that have been in serial_core.h since -pre-git history (2005) wouldn't just be getting added to the usb serial infrastructure now)...