Skip to contentSkip to conversation

2.6.0-test11 -- Failed to open /dev/ttyS0: No such device

DISCUSSION William Park 5 participants 10 messages Nov 30, 2003 – Dec 1, 2003

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Does anyone have modem working in 2.6.0-test11?

I have external modem connected to /dev/ttyS0 (COM1). Kernel 2.6.0-test11 give me
Failed to open /dev/modem: No such device
where /dev/modem is symlink to /dev/ttyS0. I've looked at /proc/interrupts and /proc/ioports, and I can't find any mention of irq=4 or io=3f8 which are the normal settings that I use.

No problem in kernel-2.4.23, though.

Signature · 2 lines
William Park, Open Geometry Consulting, <[email protected]>
Linux solution for data management and processing. 
#1 lore raw
Quoted 4 lines · William Park

Does anyone have modem working in 2.6.0-test11?

I have external modem connected to /dev/ttyS0 (COM1). Kernel 2.6.0-test11 give me

Double check your .config and attach it if in doubt.

Something like grep SERIAL .config might be enlightning.

Signature · 2 lines
http://www.PowerDNS.com      Open source, database driven DNS Software 
http://lartc.org           Linux Advanced Routing & Traffic Control HOWTO
#2 lore raw
Quoted 5 lines · bert hubert
Quoted 3 lines

Does anyone have modem working in 2.6.0-test11?
I have external modem connected to /dev/ttyS0 (COM1). Kernel 2.6.0-test11 give me

Double check your .config and attach it if in doubt. Something like grep SERIAL .config might be enlightning.

It reminds me, that I had to add serial to the list of modules loading at start to get back access to /dev/ttyS* (while upgrading from -test9 to -test10).

install serial /sbin/modprobe 8250 && { /etc/init.d/setserial modload }

Linux finwe 2.6.0-test11 #5 Fri Nov 28 01:22:33 CET 2003 i686 GNU/Linux

Gnu C                  3.3.2
Gnu make               3.80
util-linux             2.12
mount                  2.12
module-init-tools      0.9.15-pre3
e2fsprogs              1.35-WIP
jfsutils               1.1.4
xfsprogs               2.6.0
PPP                    2.4.2b3
Linux C Library        2.3.2
Dynamic linker (ldd)   2.3.2
Procps                 3.1.14
Net-tools              1.60
Console-tools          0.2.3
Sh-utils               5.0.91
Modules Loaded         isofs nls_cp852 smbfs ppp_deflate zlib_deflate zlib_inflate bsd_comp parport_pc lp parport snd_seq_midi snd_seq_oss snd_seq_midi_event snd_seq snd_pcm_oss snd_mixer_oss snd_ens1371 snd_rawmidi snd_seq_device snd_pcm snd_page_alloc snd_timer snd_ac97_codec gameport snd soundcore ppp_async ppp_generic slhc ip_nat_ftp ipt_multiport ipt_state ipt_pkttype ipt_LOG ipt_limit ipt_REJECT iptable_nat iptable_filter ip_tables ip_conntrack_ftp ip_conntrack rtc 8139too mii crc32 uhci_hcd ohci_hcd nls_iso8859_2 nls_cp437 vfat fat 8250 serial_core psmouse thermal processor fan

bye,

Signature · 1 line
Jacek Kawa  **I stepped into an avalanche,it covered up my soul [L.Cohen]**
#3 lore raw

|--==> "JK" == Jacek Kawa <[email protected]> writes:

Quoted 3 lines · JK> bert hubert wrote:
Quoted 3 lines
Quoted 3 lines

Does anyone have modem working in 2.6.0-test11?
I have external modem connected to /dev/ttyS0 (COM1). Kernel 2.6.0-test11 give me

JK> It reminds me, that I had to add serial to the list of modules JK> loading at start to get back access to /dev/ttyS* JK> (while upgrading from -test9 to -test10).

Jacek,

I _think_ this patch will bring back auto-loading of the serial module for you. Please let me know how it goes. (Bert, this won't fix your problem if you have the serial driver compiled directly into the kernel, but it might if you have it as a module.)

drivers/serial/serial_core.c +1 -0
@@ -2420,3 +2420,4 @@
 
 MODULE_DESCRIPTION("Serial driver core");
 MODULE_LICENSE("GPL");
+MODULE_ALIAS_CHARDEV(drv->major, drv->minor);
Signature · 4 lines
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
|              Ashes to ashes, dust to dust.               |
|      The proof of the pudding, is under the crust.       |
|------------------------------<[email protected]>---|
#4 lore raw
Quoted 22 lines · Steve Youngs

|--==> "JK" == Jacek Kawa <[email protected]> writes:

Quoted 3 lines · JK> bert hubert wrote:
Quoted 3 lines
Quoted 3 lines

Does anyone have modem working in 2.6.0-test11?
I have external modem connected to /dev/ttyS0 (COM1). Kernel 2.6.0-test11 give me

JK> It reminds me, that I had to add serial to the list of modules JK> loading at start to get back access to /dev/ttyS* JK> (while upgrading from -test9 to -test10).

Jacek,

I _think_ this patch will bring back auto-loading of the serial module for you. Please let me know how it goes. (Bert, this won't fix your problem if you have the serial driver compiled directly into the kernel, but it might if you have it as a module.)

drivers/serial/serial_core.c line 2423

Quoted 4 lines
drivers/serial/serial_core.c +1 -0
@@ -2420,3 +2420,4 @@
 
 MODULE_DESCRIPTION("Serial driver core");
 MODULE_LICENSE("GPL");
+MODULE_ALIAS_CHARDEV(drv->major, drv->minor);

This is wrong. serial_core should /never/ depend on a major/minor number because it doesn't know what (group of) major/minor(s) it is going to be.

The only modules which know are the hardware drivers themselves, like 8250.c.

Signature · 4 lines
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 PCMCIA      - http://pcmcia.arm.linux.org.uk/
                 2.6 Serial core
#9 lore raw
In reply to bert hubert
Quoted 9 lines · bert hubert
Quoted 4 lines · William Park

Does anyone have modem working in 2.6.0-test11?

I have external modem connected to /dev/ttyS0 (COM1). Kernel 2.6.0-test11 give me

Double check your .config and attach it if in doubt.

Something like grep SERIAL .config might be enlightning.

My apology...

CONFIG_PARPORT_SERIAL=m
# CONFIG_MOUSE_SERIAL is not set
# CONFIG_SERIAL_NONSTANDARD is not set
CONFIG_SERIAL_8250=m
CONFIG_SERIAL_8250_CS=m
CONFIG_SERIAL_8250_NR_UARTS=4
# CONFIG_SERIAL_8250_EXTENDED is not set
CONFIG_SERIAL_CORE=m
CONFIG_SND_SERIAL_U16550=m
# CONFIG_USB_SERIAL is not set

CONFIG_PPP=m
# CONFIG_PPP_MULTILINK is not set
# CONFIG_PPP_FILTER is not set
CONFIG_PPP_ASYNC=m
CONFIG_PPP_SYNC_TTY=m
CONFIG_PPP_DEFLATE=m
CONFIG_PPP_BSDCOMP=m
CONFIG_PPPOE=m

Everything related to serial and ppp are module. This is the way I have it with 2.4.23. I haven't tried any earlier version than 2.6.0-test11.

Signature · 2 lines
William Park, Open Geometry Consulting, <[email protected]>
Linux solution for data management and processing. 
#8 lore raw
In reply to Jacek Kawa
Quoted 5 lines · Jacek Kawa

It reminds me, that I had to add serial to the list of modules loading at start to get back access to /dev/ttyS* (while upgrading from -test9 to -test10).

install serial /sbin/modprobe 8250 && { /etc/init.d/setserial modload }

Yes, that did it. 'modprobe 8250' loads '8250' and 'serial_core' modules. It's odd that I have to be explicit about it in 2.6.0, whereas 2.4.23 loads 'serial' module automatically when dialing out.

Signature · 2 lines
William Park, Open Geometry Consulting, <[email protected]>
Linux solution for data management and processing. 
#5 lore raw
In reply to Steve Youngs
Quoted 11 lines · Steve Youngs
Quoted 3 lines · JK> bert hubert wrote:
Quoted 3 lines
Quoted 3 lines

Does anyone have modem working in 2.6.0-test11?
I have external modem connected to /dev/ttyS0 (COM1). Kernel 2.6.0-test11 give me

JK> It reminds me, that I had to add serial to the list of modules JK> loading at start to get back access to /dev/ttyS* JK> (while upgrading from -test9 to -test10).

I _think_ this patch will bring back auto-loading of the serial module for you. Please let me know how it goes.

Well: patched, installed new serial_core.ko, then depmod -a, and try to access ttySwhatever.

I don't see any difference... If I understood correctly, it could not load my 'fake' serial module anyway (?)

BTW. only changes,I think could affect serial directly and
made betwen test9 and test10 are:

drivers/serial/serial_core.c +3 -0
diff -Nru a/drivers/serial/serial_core.c b/drivers/serial/serial_core.c
@@ -1707,6 +1707,9 @@
 		strcat(stat_buf, "\n");
 	
 		ret += sprintf(buf + ret, stat_buf);
+	} else {
+		strcat(buf, "\n");
+		ret++;
 	}
 #undef STATBIT
 #undef INFOBIT
include/linux/serial.h +0 -1
diff -Nru a/include/linux/serial.h b/include/linux/serial.h
@@ -49,7 +49,6 @@
 	unsigned short	iomem_reg_shift;
 	unsigned int	port_high;
 	unsigned long	iomap_base;	/* cookie passed into ioremap */
-	int	reserved[1];
 };
 
 /*

I could try to narrow it down to some -bk, but it will take time...

PS. I wonder - could it be, that somewhere in the middle of
decision process test9 > test10 (as test1something)

bye

Signature · 2 lines
Jacek Kawa  **Oh, Lord, bless this thy hand grenade that with it thou
              mayest blow thy enemies to tiny bits, in thy mercy.**
#6 lore raw

|--==> "JK" == Jacek Kawa <[email protected]> writes:

Quoted 2 lines · JK> Steve Youngs wrote:
Quoted 2 lines

I _think_ this patch will bring back auto-loading of the serial module for you. Please let me know how it goes.

JK> Well: patched, installed new serial_core.ko, then depmod -a, and JK> try to access ttySwhatever.

As Russell said, the patch was wrong, revert it. Sorry about that, my bad. I really shouldn't try to hack the kernel late at night while I'm hacking other things at the same time.

I'm pretty sure that the problem has its roots in...

fs/char_dev.c +1 -1
diff -Nru a/fs/char_dev.c b/fs/char_dev.c
@@ -434,7 +434,7 @@
 
 static struct kobject *base_probe(dev_t dev, int *part, void *data)
 {
-	request_module("char-major-%d", MAJOR(dev));
+	request_module("char-major-%d-%d", MAJOR(dev), MINOR(dev));
 	return NULL;
 }
 

...from Rusty, which went into 2.6.0-test10.

Does this work any better?

drivers/serial/8250.c +2 -0
@@ -34,6 +34,7 @@
 #include <linux/serial.h>
 #include <linux/serialP.h>
 #include <linux/delay.h>
+#include <linux/device.h>
 
 #include <asm/io.h>
 #include <asm/irq.h>
@@ -2195,3 +2196,4 @@
 MODULE_PARM(force_rsa, "1-" __MODULE_STRING(PORT_RSA_MAX) "i");
 MODULE_PARM_DESC(force_rsa, "Force I/O ports for RSA");
 #endif
+MODULE_ALIAS_CHARDEV_MAJOR(TTY_MAJOR);
Signature · 4 lines
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
|              Ashes to ashes, dust to dust.               |
|      The proof of the pudding, is under the crust.       |
|------------------------------<[email protected]>---|
#7 lore raw
Quoted 8 lines · Steve Youngs
Quoted 2 lines
Quoted 2 lines

I _think_ this patch will bring back auto-loading of the serial module for you. Please let me know how it goes.

JK> Well: patched, installed new serial_core.ko, then depmod -a, and JK> try to access ttySwhatever.

As Russell said, the patch was wrong, revert it. Sorry about that, my bad.

Nothing bad happened :)

Quoted 18 lines

I'm pretty sure that the problem has its roots in...

fs/char_dev.c +1 -1
@@ -434,7 +434,7 @@
 
 static struct kobject *base_probe(dev_t dev, int *part, void *data)
 {
-	request_module("char-major-%d", MAJOR(dev));
+	request_module("char-major-%d-%d", MAJOR(dev), MINOR(dev));
 	return NULL;
 }

...from Rusty, which went into 2.6.0-test10.

Does this work any better?

drivers/serial/8250.c +0 -0

[...]

drivers/serial/8250.c line 2199

Quoted 5 lines
drivers/serial/8250.c +1 -0
@@ -2195,3 +2196,4 @@
 MODULE_PARM(force_rsa, "1-" __MODULE_STRING(PORT_RSA_MAX) "i");
 MODULE_PARM_DESC(force_rsa, "Force I/O ports for RSA");
 #endif
+MODULE_ALIAS_CHARDEV_MAJOR(TTY_MAJOR);

1. Yup, with first patch revertet everything works as before
2. With both patch applied I don't have to manually load any module (8250 and serial_core are autoloaded), but my pseudomodule serial isn't loaded anymore.

Thanks!

bye

Signature · 1 line
Jacek Kawa  **SPAM - Stowarzyszenie Polskich Artystów Muzyków**