Skip to contentSkip to conversation

serial: DCC(JTAG) serial and console emulation support

PATCH In review Daniel Walker 4 participants 44 messages Oct 7, 2010 – Oct 9, 2010 TTY layer and serial drivers

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 serial: DCC(JTAG) serial and console emulation support

Many of JTAG debuggers for ARM support DCC protocol over JTAG connection, which is very useful to debug hardwares which has no serial port. This patch adds DCC serial emulation and the console support. System timer based polling method is used for the emulation of serial input interrupt handling.

Most of the code was taken from Hyok S. Choi original work, but the inline assmebly needed some work and updating. It now supports ARMv7. Also the description above is from Hyok also.

---
 drivers/serial/Kconfig      |   31 +++
 drivers/serial/Makefile     |    1 +
 drivers/serial/dcc.c        |  432 +++++++++++++++++++++++++++++++++++++++++++
 include/linux/serial_core.h |    3 +
 4 files changed, 467 insertions(+), 0 deletions(-)
 create mode 100644 drivers/serial/dcc.c
drivers/serial/Kconfig +31 -0
index 12900f7..a54a57c 100644
@@ -495,6 +495,37 @@ config SERIAL_S3C2400
 	help
 	  Serial port support for the Samsung S3C2400 SoC
 
+config SERIAL_DCC
+	bool "JTAG ICE/ICD DCC serial port emulation support"
+	depends on ARM
+	select SERIAL_CORE
+	help
+	  This selects serial port emulation driver for ICE/ICD JTAG debugger
+	  (e.g. Trace32) for ARM architecture. You should make an terminal with
+	  DCC(JTAG1) protocol.
+
+	  if unsure, say N.
+
+config SERIAL_DCC_CONSOLE
+	bool "Support for console on JTAG ICE/ICD DCC"
+	depends on SERIAL_DCC
+	select SERIAL_CORE_CONSOLE
+	help
+	  Say Y here if you wish to use ICE/ICD JTAG DCC serial port emulation
+	  as the system console.
+
+	  if unsure, say N.
+
+config SERIAL_DCC_STDSERIAL
+	bool "Install JTAG ICE/ICD DCC as standard serial"
+	default y
+	depends on !SERIAL_8250 && SERIAL_DCC
+	help
+	  Say Y here if you want to install DCC driver as a normal serial port
+	  /dev/ttyS0 (major 4, minor 64). Otherwise, it appears as /dev/ttyJ0
+	  (major 4, minor 128) and can co-exist with other UARTs, such as
+	  8250/16C550 compatibles.
+
 config SERIAL_S3C2410
 	tristate "Samsung S3C2410 Serial port support"
 	depends on SERIAL_SAMSUNG && CPU_S3C2410
drivers/serial/Makefile +1 -0
index 1ca4fd5..896c3f6 100644
@@ -31,6 +31,7 @@ obj-$(CONFIG_SERIAL_8250_MCA) += 8250_mca.o
 obj-$(CONFIG_SERIAL_AMBA_PL010) += amba-pl010.o
 obj-$(CONFIG_SERIAL_AMBA_PL011) += amba-pl011.o
 obj-$(CONFIG_SERIAL_CLPS711X) += clps711x.o
+obj-$(CONFIG_SERIAL_DCC) += dcc.o
 obj-$(CONFIG_SERIAL_PXA) += pxa.o
 obj-$(CONFIG_SERIAL_PNX8XXX) += pnx8xxx_uart.o
 obj-$(CONFIG_SERIAL_SA1100) += sa1100.o
drivers/serial/dcc.c new file +432 -0

432 lines not on this page · raw message

include/linux/serial_core.h +3 -0
index 563e234..a360c3a 100644
@@ -196,6 +196,9 @@
 /* High Speed UART for Medfield */
 #define PORT_MFD	95
 
+/* DCC(JTAG) emulation port types */
+#define PORT_DCC_JTAG1	96
+
 #ifdef __KERNEL__
 
 #include <linux/compiler.h>
Signature · 1 line
1.7.1
#1 lore raw
Quoted 5 lines · Daniel Walker

Many of JTAG debuggers for ARM support DCC protocol over JTAG connection, which is very useful to debug hardwares which has no serial port. This patch adds DCC serial emulation and the console support. System timer based polling method is used for the emulation of serial input interrupt handling.

why use serial_core at all ? seems like you could just use the tty layer directly. i did that with drivers/char/bfin_jtag_comm.c.

although, there's the suggestion in the air that my jtag driver could be converted to the HVC framework to even further simplify things. but no one thinks it should be serial_core :).
-mike

#2 lore raw
Quoted 7 lines · Mike Frysinger
Quoted 5 lines · Daniel Walker

Many of JTAG debuggers for ARM support DCC protocol over JTAG connection, which is very useful to debug hardwares which has no serial port. This patch adds DCC serial emulation and the console support. System timer based polling method is used for the emulation of serial input interrupt handling.

Quoted 7 lines

why use serial_core at all ? seems like you could just use the tty layer directly. i did that with drivers/char/bfin_jtag_comm.c.

although, there's the suggestion in the air that my jtag driver could be converted to the HVC framework to even further simplify things. but no one thinks it should be serial_core :).
-mike

It will actually create a serial port on ttyS* if you want it to.. It could be debatable if that's useful, but I didn't rip that out because Hyok had that in his original code (he thought it was useful) ..

Daniel

Signature · 2 lines
Sent by a consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.
#3 lore raw
Quoted 11 lines · Daniel Walker
Quoted 10 lines · Mike Frysinger
Quoted 5 lines · Daniel Walker

Many of JTAG debuggers for ARM support DCC protocol over JTAG connection, which is very useful to debug hardwares which has no serial port. This patch adds DCC serial emulation and the console support. System timer based polling method is used for the emulation of serial input interrupt handling.

why use serial_core at all ?  seems like you could just use the tty layer directly.  i did that with drivers/char/bfin_jtag_comm.c.

Quoted 7 lines
Quoted 3 lines

although, there's the suggestion in the air that my jtag driver could be converted to the HVC framework to even further simplify things. but no one thinks it should be serial_core :).

It will actually create a serial port on ttyS* if you want it to.. It could be debatable if that's useful, but I didn't rip that out because Hyok had that in his original code (he thought it was useful) ..

i dont think hijacking the "ttyS" namespace is acceptable, even if it's behind a Kconfig. other people have tried in the past (myself included) and been shot down. sounds like a shim for crappy userspace apps that blindly assume /dev/ttyS*. is that really an issue anymore though now that things like /dev/ttyUSB* are so common and people are used to picking between multiple serial sources ?
-mike

#4 lore raw
Quoted 6 lines · Mike Frysinger

i dont think hijacking the "ttyS" namespace is acceptable, even if it's behind a Kconfig. other people have tried in the past (myself included) and been shot down. sounds like a shim for crappy userspace apps that blindly assume /dev/ttyS*. is that really an issue anymore though now that things like /dev/ttyUSB* are so common and people are used to picking between multiple serial sources ?

It also creates a ttyJ* which uses the same interfaces as the code for ttyS* .. I don't think allowing a Kconfig option to shim this in as a ttyS* is all that bad .. This driver is only going to be used in rare cases for debugging , having a ttyS* is just a level of flexibility ..

Daniel

Signature · 2 lines
Sent by a consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.
#5 lore raw
Quoted 8 lines · Daniel Walker
Quoted 6 lines · Mike Frysinger

i dont think hijacking the "ttyS" namespace is acceptable, even if it's behind a Kconfig.  other people have tried in the past (myself included) and been shot down.  sounds like a shim for crappy userspace apps that blindly assume /dev/ttyS*.  is that really an issue anymore though now that things like /dev/ttyUSB* are so common and people are used to picking between multiple serial sources ?

Quoted 4 lines

It also creates a ttyJ* which uses the same interfaces as the code for ttyS* .. I don't think allowing a Kconfig option to shim this in as a ttyS* is all that bad .. This driver is only going to be used in rare cases for debugging , having a ttyS* is just a level of flexibility ..

how is that any different from:
ln -s ttyJ0 /dev/ttyS0
-mike

#6 lore raw
Quoted 3 lines · Mike Frysinger

how is that any different from:
ln -s ttyJ0 /dev/ttyS0
-mike

It has the same major and minors as ttyS* does. So you don't have to run anything on the target.

Daniel

Signature · 2 lines
Sent by a consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.
#16 lore raw
In reply to Daniel Walker

drivers/serial/Kconfig config SERIAL_S3C2400 · line 528

Quoted 6 lines
drivers/serial/Kconfig +5 -0
+	  Say Y here if you want to install DCC driver as a normal serial port
+	  /dev/ttyS0 (major 4, minor 64). Otherwise, it appears as /dev/ttyJ0
+	  (major 4, minor 128) and can co-exist with other UARTs, such as
+	  8250/16C550 compatibles.
+

NAK to both

ttyJ0 is 204,186 for "JTAG1 DCC protocol based serial"

so there is an existing name and minor allocation, which as you won't be using two of them should be quite usable.

If you want to be able to switch at runtime to pretend it is ttyS0 please deal with that in your user space. The same rules apply to you as have been applied to everyone else who has tried to implement this same crap in their uart driver too (we'd have it in about 30 by now otherwise)

drivers/serial/dcc.c line 159

Quoted 5 lines
drivers/serial/dcc.c new file +4 -0
+static inline void dcc_rx_chars(struct uart_port *port)
+{
+	unsigned char ch;
+	struct tty_struct *tty = port->state->port.tty;

Not safe in the general case - you might get a hangup here then call through a NULL pointer. Please use the proper tty_port helpers.

drivers/serial/dcc.c line 183

Quoted 4 lines
drivers/serial/dcc.c new file +4 -0
+static inline void dcc_overrun_chars(struct uart_port *port)
+{
+	port->icount.overrun++;
+}

Why make this a function?

Quoted 7 lines
drivers/serial/dcc.c new file +7 -0
+static void
+dcc_set_termios(struct uart_port *port, struct ktermios *termios,
+		struct ktermios *old)
+{
+	unsigned int baud, quot;
+
+	/*

drivers/serial/dcc.c line 267

Quoted 10 lines
drivers/serial/dcc.c new file +10 -0
+	 * We don't support parity, stop bits, or anything other
+	 * than 8 bits, so clear these termios flags.
+	 */
+	termios->c_cflag &= ~(CSIZE | CSTOPB | PARENB | PARODD | CREAD);
+	termios->c_cflag |= CS8;
+
+	/*
+	 * We don't appear to support any error conditions either.
+	 */
+	termios->c_iflag &= ~(INPCK | IGNPAR | IGNBRK | BRKINT);

There is a helper function for keeping the hardware bits fixed. If your hardware is fixed then please use it.

Alan

#13 lore raw
In reply to Mike Frysinger
Quoted 9 lines · On Thu, 7 Oct 2010 15:25:34 -0400 Mike Frysinger <[email protected]> wrote:
Quoted 5 lines · Daniel Walker

Many of JTAG debuggers for ARM support DCC protocol over JTAG connection, which is very useful to debug hardwares which has no serial port. This patch adds DCC serial emulation and the console support. System timer based polling method is used for the emulation of serial input interrupt handling.

why use serial_core at all ? seems like you could just use the tty layer directly. i did that with drivers/char/bfin_jtag_comm.c.

I would agree 100% with this for the driver in question. The tty_port helpers now make it trivial to do so and it'll be small and clean as a result.

Alan

#17 lore raw
In reply to Alan Cox
Quoted 7 lines · Alan Cox

drivers/serial/Kconfig config SERIAL_S3C2400 · line 528

Quoted 5 lines
drivers/serial/Kconfig +5 -0
+	  Say Y here if you want to install DCC driver as a normal serial port
+	  /dev/ttyS0 (major 4, minor 64). Otherwise, it appears as /dev/ttyJ0
+	  (major 4, minor 128) and can co-exist with other UARTs, such as
+	  8250/16C550 compatibles.
+

NAK to both

Both what?

Quoted 9 lines

ttyJ0 is 204,186 for "JTAG1 DCC protocol based serial"

so there is an existing name and minor allocation, which as you won't be using two of them should be quite usable.

If you want to be able to switch at runtime to pretend it is ttyS0 please deal with that in your user space. The same rules apply to you as have been applied to everyone else who has tried to implement this same crap in their uart driver too (we'd have it in about 30 by now otherwise)

I would agree if this wasn't strictly for debugging embedded devices in difficult situations.. After talking to Mike, it's seems like it would be useful to have this as a ttyS* specifically because embedded devices won't always create a ttyJ* for you and ttyS* will likely already exist.

Daniel

Signature · 2 lines
Sent by a consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.
#14 lore raw
In reply to Alan Cox
Quoted 10 lines · Alan Cox
Quoted 7 lines · On Thu, 7 Oct 2010 15:25:34 -0400 Mike Frysinger <[email protected]> wrote:
Quoted 5 lines · Daniel Walker

Many of JTAG debuggers for ARM support DCC protocol over JTAG connection, which is very useful to debug hardwares which has no serial port. This patch adds DCC serial emulation and the console support. System timer based polling method is used for the emulation of serial input interrupt handling.

Quoted 6 lines
Quoted 2 lines

why use serial_core at all ? seems like you could just use the tty layer directly. i did that with drivers/char/bfin_jtag_comm.c.

I would agree 100% with this for the driver in question. The tty_port helpers now make it trivial to do so and it'll be small and clean as a result.

Can you give some example of other drivers which have done this?

Daniel

Signature · 2 lines
Sent by a consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.
#18 lore raw
In reply to Daniel Walker
Quoted 10 lines · On Thu, 07 Oct 2010 13:36:06 -0700 Daniel Walker <[email protected]> wrote:
Quoted 7 lines · Alan Cox

drivers/serial/Kconfig config SERIAL_S3C2400 · line 528

Quoted 5 lines
drivers/serial/Kconfig +5 -0
+	  Say Y here if you want to install DCC driver as a normal serial port
+	  /dev/ttyS0 (major 4, minor 64). Otherwise, it appears as /dev/ttyJ0
+	  (major 4, minor 128) and can co-exist with other UARTs, such as
+	  8250/16C550 compatibles.
+

NAK to both

Both what?

Both device allocations. Please use the 204,186 assigned for JTAG DCC.

Quoted 4 lines

I would agree if this wasn't strictly for debugging embedded devices in difficult situations.. After talking to Mike, it's seems like it would be useful to have this as a ttyS* specifically because embedded devices won't always create a ttyJ* for you and ttyS* will likely already exist.

If you are debugging an embedded device you are capable of changing it (otherwise you wouldn't debug it). Therefore the fact you might need to tweak the tty creation on the device is not a problem. If you can't change the device well there is no point debugging it is there !

We've said no over a period of about ten years to a lot of attempts to just borrow the ttyS0 range. If we'd said yes it would have been a complete mess by now.

So the answer is no.

Alan

#15 lore raw
In reply to Daniel Walker
Quoted 5 lines
Quoted 3 lines

I would agree 100% with this for the driver in question. The tty_port helpers now make it trivial to do so and it'll be small and clean as a result.

Can you give some example of other drivers which have done this?

The blackfin driver is probably the nicest example for a very simple device but see also the USB serial drivers (complex example), the SDIO serial support (shows how to do fancy hotpluggable device stuff with it).

Basically tty_port_open/tty_port_close/tty_port_hangup do all the icky nasty POSIX open/close/hangup handling and provide you with simple methods to implement activate/shutdown of a port and carrier detect etc most of which you don't even need to supply.

You can also skip a lot of other stuff like modem lines - although I can see why you might want to provide an emulated modem line response for a console emulation.

Alan

#7 lore raw
In reply to Daniel Walker
Quoted 6 lines · Daniel Walker
Quoted 2 lines · Mike Frysinger

how is that any different from:
ln -s ttyJ0 /dev/ttyS0

It has the same major and minors as ttyS* does. So you don't have to run anything on the target.

i dont see how those things are related. the major/minor are irrelevant, unless you've already hard coded these in some app that creates device nodes manually (instead of mdev/udev), but even then that's something that "needs to be run on the target". and both already have config support to transparently do something like "symlink ttyS# to XXX" as XXX is created.
-mike

#19 lore raw
In reply to Alan Cox
Quoted 21 lines · Alan Cox
Quoted 10 lines · On Thu, 07 Oct 2010 13:36:06 -0700 Daniel Walker <[email protected]> wrote:
Quoted 7 lines · Alan Cox

drivers/serial/Kconfig config SERIAL_S3C2400 · line 528

Quoted 5 lines
drivers/serial/Kconfig +5 -0
+	  Say Y here if you want to install DCC driver as a normal serial port
+	  /dev/ttyS0 (major 4, minor 64). Otherwise, it appears as /dev/ttyJ0
+	  (major 4, minor 128) and can co-exist with other UARTs, such as
+	  8250/16C550 compatibles.
+

NAK to both

Both what?

Both device allocations. Please use the 204,186 assigned for JTAG DCC.

Quoted 4 lines

I would agree if this wasn't strictly for debugging embedded devices in difficult situations.. After talking to Mike, it's seems like it would be useful to have this as a ttyS* specifically because embedded devices won't always create a ttyJ* for you and ttyS* will likely already exist.

Quoted 4 lines

If you are debugging an embedded device you are capable of changing it (otherwise you wouldn't debug it). Therefore the fact you might need to tweak the tty creation on the device is not a problem. If you can't change the device well there is no point debugging it is there !

Your making too many assumptions .. You might be able to modify the kernel, and not the userspace. So you couldn't tweak the device creation .. It's much easier in the server world ..

Hyok should really be addressing this .. If he added for some random useless reason , then for sure we can rip it out .. Otherwise I think it deserves some more discussion ..

Quoted 5 lines

We've said no over a period of about ten years to a lot of attempts to just borrow the ttyS0 range. If we'd said yes it would have been a complete mess by now.

So the answer is no.

Nothing can be unilateral, there's always room for exceptions. You should say something more like "it's possible, but unlikely".

Daniel

Signature · 2 lines
Sent by a consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.
#8 lore raw
In reply to Mike Frysinger
Quoted 8 lines · Mike Frysinger
Quoted 6 lines · Daniel Walker
Quoted 2 lines · Mike Frysinger

how is that any different from:
ln -s ttyJ0 /dev/ttyS0

It has the same major and minors as ttyS* does. So you don't have to run anything on the target.

Quoted 6 lines

i dont see how those things are related. the major/minor are irrelevant, unless you've already hard coded these in some app that creates device nodes manually (instead of mdev/udev), but even then that's something that "needs to be run on the target". and both already have config support to transparently do something like "symlink ttyS# to XXX" as XXX is created.

Something does ultimately run on all targets, the issues is does the person debugging have the ability to change it .. It's not always the case that they can change it. The userspace may not have udev at all. The userspace may consist of a single binary .. I'm not making any assumptions about the userspace here, it can be anything.

Daniel

Signature · 2 lines
Sent by a consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.
#20 lore raw
In reply to Daniel Walker
Quoted 16 lines · Daniel Walker
Quoted 15 lines · Alan Cox
Quoted 13 lines · On Thu, 07 Oct 2010 13:36:06 -0700 Daniel Walker wrote:
Quoted 7 lines · Alan Cox
Quoted 5 lines

+         Say Y here if you want to install DCC driver as a normal serial port
+         /dev/ttyS0 (major 4, minor 64). Otherwise, it appears as /dev/ttyJ0
+         (major 4, minor 128) and can co-exist with other UARTs, such as
+         8250/16C550 compatibles.
+

NAK to both

I would agree if this wasn't strictly for debugging embedded devices in difficult situations.. After talking to Mike, it's seems like it would be useful to have this as a ttyS* specifically because embedded devices won't always create a ttyJ* for you and ttyS* will likely already exist.

Quoted 8 lines
Quoted 4 lines

If you are debugging an embedded device you are capable of changing it (otherwise you wouldn't debug it). Therefore the fact you might need to tweak the tty creation on the device is not a problem. If you can't change the device well there is no point debugging it is there !

Your making too many assumptions .. You might be able to modify the kernel, and not the userspace. So you couldn't tweak the device creation .. It's much easier in the server world ..

this driver isnt in the kernel. there is no ABI to preserve. if people want to use the mainline driver, they can fix their userspace. i dont think the "binary userspace" argument is terribly relevant to new driver submissions.
-mike

#9 lore raw
In reply to Daniel Walker
Quoted 16 lines · Daniel Walker
Quoted 8 lines · Mike Frysinger
Quoted 6 lines · Daniel Walker
Quoted 2 lines · Mike Frysinger

how is that any different from:
ln -s ttyJ0 /dev/ttyS0

It has the same major and minors as ttyS* does. So you don't have to run anything on the target.

Quoted 6 lines

i dont see how those things are related.  the major/minor are irrelevant, unless you've already hard coded these in some app that creates device nodes manually (instead of mdev/udev), but even then that's something that "needs to be run on the target".  and both already have config support to transparently do something like "symlink ttyS# to XXX" as XXX is created.

Quoted 5 lines

Something does ultimately run on all targets, the issues is does the person debugging have the ability to change it .. It's not always the case that they can change it. The userspace may not have udev at all. The userspace may consist of a single binary .. I'm not making any assumptions about the userspace here, it can be anything.

then there is already a static rootfs with pre-populated /dev which would be fixed, or the single binary fixed. if the options are "my system works" or "my system doesnt work", i cant imagine people would continue to attempt to use "/dev/ttyS#" if it didnt exist or didnt work.
-mike

#35 lore raw
In reply to Daniel Walker
Quoted 3 lines

Your making too many assumptions .. You might be able to modify the kernel, and not the userspace. So you couldn't tweak the device creation .. It's much easier in the server world ..

Spare me the embedded is different lecture. It's not. In fact it's usually easier because the environment you are testing is usually standalone, doesn't cause a million dollar an hour downtime if you get it wrong and was designed for debug/test so actually has a jtag port.

If you can modify the file system you can make /dev/ttyS0 the major/minor of the jtag port.

If you can load anything in front of the userspace you can rename the device/move it

So I find your example case fictional. In fact the only time I can see you having a box so tightly locked up you can only tweak the kernel is jailbreaking it, in which case you are such a fringe usage case and technically competent that you can just do custom patches as you'll be doing anyway for other bits.

Quoted 8 lines
Quoted 5 lines

We've said no over a period of about ten years to a lot of attempts to just borrow the ttyS0 range. If we'd said yes it would have been a complete mess by now.

So the answer is no.

Nothing can be unilateral, there's always room for exceptions. You should say something more like "it's possible, but unlikely".

This argument gets regurgitated every so often and nobody has yet provided a plausable reason to make a nasty mess.

Either way I would suggest the path forward is

- Look at the blackfin jtag using tty_port alone and see if it looks cleaner
- Fix the bugs noted
- Submit a driver that uses the existing allocated jtag major/minor only

and then have a debate about ttyS0 hackery separately.

Alan

#21 lore raw
In reply to Mike Frysinger
Quoted 22 lines · Mike Frysinger
Quoted 21 lines · Daniel Walker
Quoted 19 lines · Alan Cox
Quoted 7 lines · On Thu, 07 Oct 2010 13:36:06 -0700 Daniel Walker wrote:
Quoted 6 lines · Alan Cox
Quoted 5 lines
drivers/serial/Kconfig +5 -0
+         Say Y here if you want to install DCC driver as a normal serial port
+         /dev/ttyS0 (major 4, minor 64). Otherwise, it appears as /dev/ttyJ0
+         (major 4, minor 128) and can co-exist with other UARTs, such as
+         8250/16C550 compatibles.
+
Quoted 6 lines
Quoted 1 line

NAK to both

I would agree if this wasn't strictly for debugging embedded devices in difficult situations.. After talking to Mike, it's seems like it would be useful to have this as a ttyS* specifically because embedded devices won't always create a ttyJ* for you and ttyS* will likely already exist.

If you are debugging an embedded device you are capable of changing it (otherwise you wouldn't debug it). Therefore the fact you might need to tweak the tty creation on the device is not a problem. If you can't change the device well there is no point debugging it is there !

Quoted 8 lines
Quoted 3 lines

Your making too many assumptions .. You might be able to modify the kernel, and not the userspace. So you couldn't tweak the device creation .. It's much easier in the server world ..

this driver isnt in the kernel. there is no ABI to preserve. if people want to use the mainline driver, they can fix their userspace. i dont think the "binary userspace" argument is terribly relevant to new driver submissions.

It is , what's to stop any userspace from using ttyS0 ? .. There's lots of binary userspaces in the embedded world, along with many strange usages ..

Daniel

Signature · 2 lines
Sent by a consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.
#39 lore raw
In reply to Daniel Walker
Quoted 3 lines · Daniel Walker

Your making too many assumptions .. You might be able to modify the kernel, and not the userspace. So you couldn't tweak the device creation .. It's much easier in the server world ..

You're saying it's easier to replace an embedded kernel than a userspace file on an embedded system? Heh, that's funny.

And no, it's not easier to repace a kernel on a "server" than a userspace file, due to all sorts of good reasons (certifications, support contracts, etc...)

Quoted 8 lines
Quoted 5 lines

We've said no over a period of about ten years to a lot of attempts to just borrow the ttyS0 range. If we'd said yes it would have been a complete mess by now.

So the answer is no.

Nothing can be unilateral, there's always room for exceptions. You should say something more like "it's possible, but unlikely".

Hm, how about this, as the TTY and serial driver[1] maintainer, I will not accept this kind of patch at all.

Is that final enough for you?

thanks,

greg k-h

[1] http://www.kernel.org/pub/linux/kernel/people/gregkh/gregkh-2.6/gregkh-06-tty/serial-mark-the-8250-driver-as-maintained.patch

#10 lore raw
In reply to Mike Frysinger
Quoted 23 lines · Mike Frysinger
Quoted 16 lines · Daniel Walker
Quoted 8 lines · Mike Frysinger
Quoted 6 lines · Daniel Walker
Quoted 2 lines · Mike Frysinger

how is that any different from:
ln -s ttyJ0 /dev/ttyS0

It has the same major and minors as ttyS* does. So you don't have to run anything on the target.

Quoted 6 lines

i dont see how those things are related. the major/minor are irrelevant, unless you've already hard coded these in some app that creates device nodes manually (instead of mdev/udev), but even then that's something that "needs to be run on the target". and both already have config support to transparently do something like "symlink ttyS# to XXX" as XXX is created.

Quoted 5 lines

Something does ultimately run on all targets, the issues is does the person debugging have the ability to change it .. It's not always the case that they can change it. The userspace may not have udev at all. The userspace may consist of a single binary .. I'm not making any assumptions about the userspace here, it can be anything.

Quoted 5 lines

then there is already a static rootfs with pre-populated /dev which would be fixed, or the single binary fixed. if the options are "my system works" or "my system doesnt work", i cant imagine people would continue to attempt to use "/dev/ttyS#" if it didnt exist or didnt work.

Ideally you would want this driver to work in any situation .. If it's setup to use ttyS0 for debugging, even if it doesn't exist, then this driver would be able to stand in for that interface.

Daniel

Signature · 2 lines
Sent by a consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.
#11 lore raw
Quoted 30 lines · Daniel Walker
Quoted 23 lines · Mike Frysinger
Quoted 16 lines · Daniel Walker
Quoted 8 lines · Mike Frysinger
Quoted 6 lines · Daniel Walker
Quoted 2 lines · Mike Frysinger

how is that any different from:
ln -s ttyJ0 /dev/ttyS0

It has the same major and minors as ttyS* does. So you don't have to run anything on the target.

Quoted 6 lines

i dont see how those things are related.  the major/minor are irrelevant, unless you've already hard coded these in some app that creates device nodes manually (instead of mdev/udev), but even then that's something that "needs to be run on the target".  and both already have config support to transparently do something like "symlink ttyS# to XXX" as XXX is created.

Quoted 5 lines

Something does ultimately run on all targets, the issues is does the person debugging have the ability to change it .. It's not always the case that they can change it. The userspace may not have udev at all. The userspace may consist of a single binary .. I'm not making any assumptions about the userspace here, it can be anything.

Quoted 5 lines

then there is already a static rootfs with pre-populated /dev which would be fixed, or the single binary fixed.  if the options are "my system works" or "my system doesnt work", i cant imagine people would continue to attempt to use "/dev/ttyS#" if it didnt exist or didnt work.

Quoted 3 lines

Ideally you would want this driver to work in any situation .. If it's setup to use ttyS0 for debugging, even if it doesn't exist, then this driver would be able to stand in for that interface.

i dont think that's the case. "any situation" is way too vague and invites mounds of crap to be included in the kernel which really should be in userspace. ive never had a problem with my embedded work using ttyBF# or ttyBFJC# or ttySS# for the Blackfin serial devices, nor have i heard customers complain that the file absolutely must be named "ttyS#". ive found that simply informing them "to use ttyBF#" has always been sufficient.
-mike

#36 lore raw
In reply to Alan Cox
Quoted 8 lines · Alan Cox
Quoted 3 lines

Your making too many assumptions .. You might be able to modify the kernel, and not the userspace. So you couldn't tweak the device creation .. It's much easier in the server world ..

Spare me the embedded is different lecture. It's not. In fact it's usually easier because the environment you are testing is usually standalone, doesn't cause a million dollar an hour downtime if you get it wrong and was designed for debug/test so actually has a jtag port.

fair enough, it's hard for both of us.

Quoted 11 lines

If you can modify the file system you can make /dev/ttyS0 the major/minor of the jtag port.

If you can load anything in front of the userspace you can rename the device/move it

So I find your example case fictional. In fact the only time I can see you having a box so tightly locked up you can only tweak the kernel is jailbreaking it, in which case you are such a fringe usage case and technically competent that you can just do custom patches as you'll be doing anyway for other bits.

There's lots of other situations that can come up. You can't in anyway say that there is some certain set of situations where you'll always have X, Y, and Z.. There have been enough strange situations that I don't think it's right for you or me to assume we just know what they all are.

Quoted 11 lines
Quoted 8 lines
Quoted 5 lines

We've said no over a period of about ten years to a lot of attempts to just borrow the ttyS0 range. If we'd said yes it would have been a complete mess by now.

So the answer is no.

Nothing can be unilateral, there's always room for exceptions. You should say something more like "it's possible, but unlikely".

This argument gets regurgitated every so often and nobody has yet provided a plausable reason to make a nasty mess.

There's plenty of nasty messes in the kernel, "they" must have had some justification ..

Quoted 4 lines

Either way I would suggest the path forward is

- Look at the blackfin jtag using tty_port alone and see if it looks cleaner

Yeah ..

Quoted 1 line

- Fix the bugs noted

Check,

Quoted 1 line

- Submit a driver that uses the existing allocated jtag major/minor only

The driver does this already. It uses ttyJ* by default, and ttyS is an option ..

Quoted 1 line

and then have a debate about ttyS0 hackery separately.

Fine with me..

Signature · 2 lines
Sent by a consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.
#40 lore raw
In reply to Greg KH
Quoted 7 lines · Greg KH
Quoted 3 lines · Daniel Walker

Your making too many assumptions .. You might be able to modify the kernel, and not the userspace. So you couldn't tweak the device creation .. It's much easier in the server world ..

You're saying it's easier to replace an embedded kernel than a userspace file on an embedded system? Heh, that's funny.

I'm saying it _can_ happen that it's easier to replace the kernel.. I'm not saying it's always the case. We're talking about debugging situations in an embedded environment which don't always follow a normal path.

Quoted 6 lines
Quoted 6 lines
Quoted 5 lines

We've said no over a period of about ten years to a lot of attempts to just borrow the ttyS0 range. If we'd said yes it would have been a complete mess by now.

So the answer is no.

Quoted 7 lines
Quoted 2 lines

Nothing can be unilateral, there's always room for exceptions. You should say something more like "it's possible, but unlikely".

Hm, how about this, as the TTY and serial driver[1] maintainer, I will not accept this kind of patch at all.

Is that final enough for you?

So you don't like it, that's fair enough .. <thinks>I wonder what other maintainers I can send this too</thinks> ;)

Can you be more specific about your objections .. The discussion over the ttyS* thing is still happening. I'm waiting for Hyok to give a good reason for why he wrote that part of it. I can imagine good reasons for why that part would exist, which is what I'm discussing with Alan and Mike.

Daniel

Signature · 2 lines
Sent by a consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.
#12 lore raw
In reply to Mike Frysinger
Quoted 4 lines · Mike Frysinger
Quoted 3 lines

Ideally you would want this driver to work in any situation .. If it's setup to use ttyS0 for debugging, even if it doesn't exist, then this driver would be able to stand in for that interface.

Quoted 8 lines

i dont think that's the case. "any situation" is way too vague and invites mounds of crap to be included in the kernel which really should be in userspace. ive never had a problem with my embedded work using ttyBF# or ttyBFJC# or ttySS# for the Blackfin serial devices, nor have i heard customers complain that the file absolutely must be named "ttyS#". ive found that simply informing them "to use ttyBF#" has always been sufficient.
-mike

It's kind of a "your milage may vary" situation .. From my perspective it doesn't hurt anything to have an exception for this, considering what it is used for.

Daniel

Signature · 2 lines
Sent by a consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.
#41 lore raw
In reply to Daniel Walker
Quoted 6 lines · Daniel Walker
Quoted 5 lines · Greg KH
Quoted 3 lines · Daniel Walker

Your making too many assumptions .. You might be able to modify the kernel, and not the userspace. So you couldn't tweak the device creation .. It's much easier in the server world ..

Quoted 7 lines
Quoted 2 lines

You're saying it's easier to replace an embedded kernel than a userspace file on an embedded system? Heh, that's funny.

I'm saying it _can_ happen that it's easier to replace the kernel.. I'm not saying it's always the case. We're talking about debugging situations in an embedded environment which don't always follow a normal path.

Nothing is "normal", so stop trying to say it is easier in the "server" world please.

Quoted 12 lines
Quoted 12 lines
Quoted 8 lines
Quoted 5 lines

We've said no over a period of about ten years to a lot of attempts to just borrow the ttyS0 range. If we'd said yes it would have been a complete mess by now.

So the answer is no.

Nothing can be unilateral, there's always room for exceptions. You should say something more like "it's possible, but unlikely".

Hm, how about this, as the TTY and serial driver[1] maintainer, I will not accept this kind of patch at all.

Quoted 6 lines
Quoted 1 line

Is that final enough for you?

So you don't like it, that's fair enough .. <thinks>I wonder what other maintainers I can send this too</thinks> ;)

Can you be more specific about your objections

See everything that Mike and Alan wrote, do I need to repeat it?

greg k-h

#42 lore raw
Quoted 10 lines · Greg KH
Quoted 9 lines · Daniel Walker
Quoted 7 lines · Greg KH
Quoted 3 lines · Daniel Walker

Your making too many assumptions .. You might be able to modify the kernel, and not the userspace. So you couldn't tweak the device creation .. It's much easier in the server world ..

You're saying it's easier to replace an embedded kernel than a userspace file on an embedded system? Heh, that's funny.

Quoted 7 lines
Quoted 4 lines

I'm saying it _can_ happen that it's easier to replace the kernel.. I'm not saying it's always the case. We're talking about debugging situations in an embedded environment which don't always follow a normal path.

Nothing is "normal", so stop trying to say it is easier in the "server" world please.

Ok ..

Quoted 12 lines
Quoted 12 lines
Quoted 12 lines
Quoted 8 lines
Quoted 5 lines

We've said no over a period of about ten years to a lot of attempts to just borrow the ttyS0 range. If we'd said yes it would have been a complete mess by now.

So the answer is no.

Nothing can be unilateral, there's always room for exceptions. You should say something more like "it's possible, but unlikely".

Hm, how about this, as the TTY and serial driver[1] maintainer, I will not accept this kind of patch at all.

Quoted 8 lines
Quoted 6 lines
Quoted 1 line

Is that final enough for you?

So you don't like it, that's fair enough .. <thinks>I wonder what other maintainers I can send this too</thinks> ;)

Can you be more specific about your objections

See everything that Mike and Alan wrote, do I need to repeat it?

I'm not sure .. I would say "Yes" if I didn't get the feeling your already to tear my head off.

I think the only disputed issue is the use of ttyS as options part of the driver ..

Daniel

Signature · 2 lines
Sent by a consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.
#43 lore raw
Quoted 23 lines · Daniel Walker
Quoted 22 lines · Greg KH
Quoted 20 lines · Daniel Walker
Quoted 7 lines · Greg KH
Quoted 6 lines · Daniel Walker
Quoted 5 lines

We've said no over a period of about ten years to a lot of attempts to just borrow the ttyS0 range. If we'd said yes it would have been a complete mess by now.

So the answer is no.

Quoted 7 lines
Quoted 2 lines

Nothing can be unilateral, there's always room for exceptions. You should say something more like "it's possible, but unlikely".

Hm, how about this, as the TTY and serial driver[1] maintainer, I will not accept this kind of patch at all.

Is that final enough for you?

So you don't like it, that's fair enough .. <thinks>I wonder what other maintainers I can send this too</thinks> ;)

Can you be more specific about your objections

Quoted 7 lines
Quoted 1 line

See everything that Mike and Alan wrote, do I need to repeat it?

I'm not sure .. I would say "Yes" if I didn't get the feeling your already to tear my head off.

I think the only disputed issue is the use of ttyS as options part of the driver ..

and using "serial_core" at all in favor of HVC or tty_core (the latter might be easier for you as the former is a bit of unknown atm) -mike

#22 lore raw
In reply to Daniel Walker
Quoted 3 lines

It is , what's to stop any userspace from using ttyS0 ? .. There's lots of binary userspaces in the embedded world, along with many strange usages ..

So you change your file system, whoopee.

And if your hypothetical user isn't able to do this then maybe instead of trying to screw up the kernel for everyone they should ask an undergraduate student who is smart enough to help them.

Alan

#37 lore raw
In reply to Daniel Walker
Quoted 4 lines
Quoted 1 line

- Submit a driver that uses the existing allocated jtag major/minor only

The driver does this already. It uses ttyJ* by default, and ttyS is an option ..

It didn't seem to be using the correct major/minor however ?

#38 lore raw
Quoted 6 lines · Alan Cox
Quoted 4 lines
Quoted 1 line

- Submit a driver that uses the existing allocated jtag major/minor only

The driver does this already. It uses ttyJ* by default, and ttyS is an option ..

It didn't seem to be using the correct major/minor however ?

yeah, it looks like it. ttyJ was registered specifically for this driver..

Daniel

Signature · 3 lines
Sent by an consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora
Forum.
#23 lore raw
In reply to Alan Cox
Quoted 5 lines · Alan Cox
Quoted 3 lines

It is , what's to stop any userspace from using ttyS0 ? .. There's lots of binary userspaces in the embedded world, along with many strange usages ..

So you change your file system, whoopee.

I'm assuming Hyok added this because it wasn't easy or possible to change the file system .. I can imagine there are debugging situations where that's true.

Quoted 3 lines

And if your hypothetical user isn't able to do this then maybe instead of trying to screw up the kernel for everyone they should ask an undergraduate student who is smart enough to help them.

I don't think the kernel is going to implode if we allow an optional ttyS override for debugging purposes.. I just don't see that "screwing" up the kernel. I can even embedded that inside the C file, and not make it a Kconfig option ..

Daniel

Signature · 3 lines
Sent by an consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora
Forum.
#24 lore raw
Quoted 8 lines · Daniel Walker
Quoted 3 lines

And if your hypothetical user isn't able to do this then maybe instead of trying to screw up the kernel for everyone they should ask an undergraduate student who is smart enough to help them.

I don't think the kernel is going to implode if we allow an optional ttyS override for debugging purposes.. I just don't see that "screwing" up the kernel. I can even embedded that inside the C file, and not make it a Kconfig option ..

I admire your persistence, but perhaps you should take a step back and look at the history here. You just want this override. Then someone else does, and then, someone else. Eventually we are all overridden.

Remember, you aren't unique here, as much as you might feel like you are :)

Please listen to what Alan and Mike are saying, experience matters, and they are trying to impart it to you, despite your most fervent attempts to avoid it.

good luck,

greg k-h

#25 lore raw
Quoted 10 lines · Greg KH
Quoted 8 lines · Daniel Walker
Quoted 3 lines

And if your hypothetical user isn't able to do this then maybe instead of trying to screw up the kernel for everyone they should ask an undergraduate student who is smart enough to help them.

I don't think the kernel is going to implode if we allow an optional ttyS override for debugging purposes.. I just don't see that "screwing" up the kernel. I can even embedded that inside the C file, and not make it a Kconfig option ..

Quoted 6 lines

I admire your persistence, but perhaps you should take a step back and look at the history here. You just want this override. Then someone else does, and then, someone else. Eventually we are all overridden.

Remember, you aren't unique here, as much as you might feel like you are :)

It's not that _I_ want it .. If you look at the code it's not that huge a part of the source .. It's got it's own ttyJ already, and it's ifdefe'd. It would be pretty easy to remove it. The other thing is that it's not something I use, so for me it's kind of take it or leave it.

So why am I arguing to keep it? Well the original author added it for a purpose, which I can't assume was useless .. I can also imagine situations when it would be helpful to do this.

To me it's more a question of if the override is useful or not .. If it's useful then it's just about organizing the code so the kernel doesn't melt down if we allow it.

Thinking about it overrides are actually allowed right now, it's just that you have to mod the source some to do it.

Quoted 3 lines

Please listen to what Alan and Mike are saying, experience matters, and they are trying to impart it to you, despite your most fervent attempts to avoid it.

The whole "fervent" thing is kind of funny to me.. I'm just persistent, that's what it's comes down to. Like I said above I'm not "in love" with this feature. I am actually taking Mike and Alan's advice by the way, so it's not like I'm not listening (or reading in this case)..

Daniel

Signature · 3 lines
Sent by an consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora
Forum.
#26 lore raw
In reply to Daniel Walker
Quoted 3 lines

I'm assuming Hyok added this because it wasn't easy or possible to change the file system .. I can imagine there are debugging situations where that's true.

Well stop assuming. You cc'd Hyok so let him speak for himself if he cares about it and has an example.

Quoted 3 lines

I don't think the kernel is going to implode if we allow an optional ttyS override for debugging purposes.. I just don't see that "screwing" up the kernel.

That is perhaps why you are not a subsystem maintainer.

Alan

#27 lore raw
Quoted 6 lines · Alan Cox
Quoted 3 lines

I'm assuming Hyok added this because it wasn't easy or possible to change the file system .. I can imagine there are debugging situations where that's true.

Well stop assuming. You cc'd Hyok so let him speak for himself if he cares about it and has an example.

I've been waiting, but he's not lightning fast responding it seems.

Quoted 5 lines
Quoted 3 lines

I don't think the kernel is going to implode if we allow an optional ttyS override for debugging purposes.. I just don't see that "screwing" up the kernel.

That is perhaps why you are not a subsystem maintainer.

No, I'm sure it's not.

What I was telling Greg is that this type of override is allowed right now, if you mod the source (everything is allowed if you mod the source right?) ..

Daniel

Signature · 3 lines
Sent by an consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora
Forum.
#28 lore raw
Quoted 5 lines · Daniel Walker
Quoted 4 lines · Alan Cox
Quoted 3 lines

I don't think the kernel is going to implode if we allow an optional ttyS override for debugging purposes.. I just don't see that "screwing" up the kernel.

Quoted 7 lines
Quoted 1 line

That is perhaps why you are not a subsystem maintainer.

No, I'm sure it's not.

What I was telling Greg is that this type of override is allowed right now, if you mod the source (everything is allowed if you mod the source right?) ..

just because someone can mod the source code to add a root shell doesnt mean we should accept patches to do it
-mike

#29 lore raw
Quoted 8 lines · Mike Frysinger
Quoted 7 lines · Daniel Walker
Quoted 5 lines · Alan Cox
Quoted 3 lines

I don't think the kernel is going to implode if we allow an optional ttyS override for debugging purposes.. I just don't see that "screwing" up the kernel.

That is perhaps why you are not a subsystem maintainer.

Quoted 8 lines
Quoted 5 lines

No, I'm sure it's not.

What I was telling Greg is that this type of override is allowed right now, if you mod the source (everything is allowed if you mod the source right?) ..

just because someone can mod the source code to add a root shell doesnt mean we should accept patches to do it

That's not what I was getting at. If you can mod the source to add the override, then adding in a patch which also requires you to mod the source to add the override should be acceptable because it's no different that what we currently have.

Daniel

Signature · 3 lines
Sent by an consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora
Forum.
#30 lore raw
Quoted 15 lines · Daniel Walker
Quoted 14 lines · Mike Frysinger
Quoted 12 lines · Daniel Walker
Quoted 5 lines · Alan Cox
Quoted 3 lines

I don't think the kernel is going to implode if we allow an optional ttyS override for debugging purposes.. I just don't see that "screwing" up the kernel.

That is perhaps why you are not a subsystem maintainer.

No, I'm sure it's not.

What I was telling Greg is that this type of override is allowed right now, if you mod the source (everything is allowed if you mod the source right?) ..

Quoted 7 lines
Quoted 2 lines

just because someone can mod the source code to add a root shell doesnt mean we should accept patches to do it

That's not what I was getting at. If you can mod the source to add the override, then adding in a patch which also requires you to mod the source to add the override should be acceptable because it's no different that what we currently have.

sorry, but you lost me. all i see is "drop the hack code". -mike

#31 lore raw
Quoted 19 lines · Mike Frysinger
Quoted 18 lines · Daniel Walker
Quoted 16 lines · Mike Frysinger
Quoted 5 lines · Daniel Walker
Quoted 4 lines · Alan Cox
Quoted 3 lines

I don't think the kernel is going to implode if we allow an optional ttyS override for debugging purposes.. I just don't see that "screwing" up the kernel.

Quoted 7 lines
Quoted 1 line

That is perhaps why you are not a subsystem maintainer.

No, I'm sure it's not.

What I was telling Greg is that this type of override is allowed right now, if you mod the source (everything is allowed if you mod the source right?) ..

just because someone can mod the source code to add a root shell doesnt mean we should accept patches to do it

Quoted 6 lines
Quoted 4 lines

That's not what I was getting at. If you can mod the source to add the override, then adding in a patch which also requires you to mod the source to add the override should be acceptable because it's no different that what we currently have.

sorry, but you lost me. all i see is "drop the hack code".

I can make it simpler for you. You add code into Linux, people can modify it, and you can't control that.

Daniel

Signature · 3 lines
Sent by an consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora
Forum.
#32 lore raw
Quoted 25 lines · Daniel Walker
Quoted 24 lines · Mike Frysinger
Quoted 22 lines · Daniel Walker
Quoted 8 lines · Mike Frysinger
Quoted 7 lines · Daniel Walker
Quoted 5 lines · Alan Cox
Quoted 3 lines

I don't think the kernel is going to implode if we allow an optional ttyS override for debugging purposes.. I just don't see that "screwing" up the kernel.

That is perhaps why you are not a subsystem maintainer.

Quoted 8 lines
Quoted 5 lines

No, I'm sure it's not.

What I was telling Greg is that this type of override is allowed right now, if you mod the source (everything is allowed if you mod the source right?) ..

just because someone can mod the source code to add a root shell doesnt mean we should accept patches to do it

That's not what I was getting at. If you can mod the source to add the override, then adding in a patch which also requires you to mod the source to add the override should be acceptable because it's no different that what we currently have.

Quoted 4 lines
Quoted 1 line

sorry, but you lost me.  all i see is "drop the hack code".

I can make it simpler for you. You add code into Linux, people can modify it, and you can't control that.

while true, that lacks justification for merging hacks into mainline -mike

#33 lore raw
Quoted 26 lines · Mike Frysinger
Quoted 25 lines · Daniel Walker
Quoted 24 lines · Mike Frysinger
Quoted 22 lines · Daniel Walker
Quoted 8 lines · Mike Frysinger
Quoted 7 lines · Daniel Walker
Quoted 5 lines · Alan Cox
Quoted 3 lines

I don't think the kernel is going to implode if we allow an optional ttyS override for debugging purposes.. I just don't see that "screwing" up the kernel.

That is perhaps why you are not a subsystem maintainer.

Quoted 8 lines
Quoted 5 lines

No, I'm sure it's not.

What I was telling Greg is that this type of override is allowed right now, if you mod the source (everything is allowed if you mod the source right?) ..

just because someone can mod the source code to add a root shell doesnt mean we should accept patches to do it

That's not what I was getting at. If you can mod the source to add the override, then adding in a patch which also requires you to mod the source to add the override should be acceptable because it's no different that what we currently have.

Quoted 6 lines
Quoted 4 lines
Quoted 1 line

sorry, but you lost me. all i see is "drop the hack code".

I can make it simpler for you. You add code into Linux, people can modify it, and you can't control that.

while true, that lacks justification for merging hacks into mainline

No one is merging hack, or giving justification for that.. So not sure why your going down that thread again .

Daniel

Signature · 3 lines
Sent by an consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora
Forum.
#34 lore raw
Quoted 29 lines · Daniel Walker
Quoted 28 lines · Mike Frysinger
Quoted 27 lines · Daniel Walker
Quoted 25 lines · Mike Frysinger
Quoted 15 lines · Daniel Walker
Quoted 14 lines · Mike Frysinger
Quoted 12 lines · Daniel Walker
Quoted 5 lines · Alan Cox
Quoted 3 lines

I don't think the kernel is going to implode if we allow an optional ttyS override for debugging purposes.. I just don't see that "screwing" up the kernel.

That is perhaps why you are not a subsystem maintainer.

No, I'm sure it's not.

What I was telling Greg is that this type of override is allowed right now, if you mod the source (everything is allowed if you mod the source right?) ..

Quoted 7 lines
Quoted 2 lines

just because someone can mod the source code to add a root shell doesnt mean we should accept patches to do it

That's not what I was getting at. If you can mod the source to add the override, then adding in a patch which also requires you to mod the source to add the override should be acceptable because it's no different that what we currently have.

sorry, but you lost me.  all i see is "drop the hack code".

Quoted 7 lines
Quoted 4 lines
Quoted 2 lines

I can make it simpler for you. You add code into Linux, people can modify it, and you can't control that.

while true, that lacks justification for merging hacks into mainline

No one is merging hack, or giving justification for that.. So not sure why your going down that thread again .

forget it; i'm completely bored with this whole driver now. people have been clear on what it will take for this to be merged, so ball is in your court now.
-mike