After installing any serial-5.00 or later in kernels 2.2.18, 19 or 20,
rs_init is entered twice. This results in the serial timer being added
twice, which corrupts the timer list and causes an infinite loop. The
symptom is a complete hang on boot, usually just after the "Freeing
unused kernel memory" message.
Pre 18 the call path is do_basic_setup() -> device_setup() ->
chr_dev_setup() -> tty_init() -> rs_init().
In 2.4.x rs_init is entered with the module_init() (__initcall) magic.
The attached is for 2.2.20, and simply removes the tty_init() ->
rs_init() path. This mimics 2.4.x.
There might be an issue; do_initcalls() is slightly later in
do_basic_setup() than device_setup(), but the in between code looks
similar to 2.4.x kernels so I don't think it's a problem delaying the
serial init.
I'll post the slightly less elegant hack to the serial driver that
compensates for the broken kernels shortly.
Patch status is: Works For Me. Testers needed.
..Stu
----diff -ru linux-2.2.20/drivers/char/tty_io.c linux-2.2.20-serial-init-fix/drivers/char/tty_io.c --- linux-2.2.20/drivers/char/tty_io.c Fri Nov 2 11:39:06 2001 +++ linux-2.2.20-serial-init-fix/drivers/char/tty_io.c Mon Nov 5 22:03:54 2001 @@ -2189,9 +2189,6 @@ #ifdef CONFIG_ESPSERIAL /* init ESP before rs, so rs doesn't see the port */ espserial_init(); #endif -#ifdef CONFIG_SERIAL - rs_init(); -#endif #ifdef CONFIG_COMPUTONE ip2_init(); #endif
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Wed Nov 07 2001 - 21:00:27 EST