2002-10-17 17:17:21

by Hiroshi Miura

Subject: APM work around for bad bios.

Hello,

I use CASIO CASSIOPEIA FIVA 101 and 103. (http://www.da-cha.org/fiva/fiva.html)
this use Cyrix MediaGX and Award BIOS.
This APM BIOS report broken value for dseg_len.
it asumes granularity is 1.

I made a work around for this situation.

* if (cseg_len < bios.offset) BIOS report BAD len value.
-- segment length must be always larger than code offset

* if (dseg_len <= 0x40 ) BIOS asumes granularity =1.
-- 0x40 * 4kB = 64kB, my pc reports 0x40.


diff -urB -x .config -x '*.[oasS]' -x '*.in' -x '*.rej' -x '*.orig' linux-2.5.43-orig/arch/i386/kernel/apm.c linux-2.5.43/arch/i386/kernel/apm.c
--- linux-2.5.43-orig/arch/i386/kernel/apm.c 2002-10-12 13:21:05.000000000 +0900
+++ linux-2.5.43/arch/i386/kernel/apm.c 2002-10-14 21:36:14.000000000 +0900
@@ -1980,6 +2141,14 @@
(apm_info.bios.cseg_16_len - 1) & 0xffff);
_set_limit((char *)&cpu_gdt_table[i][APM_DS >> 3],
(apm_info.bios.dseg_len - 1) & 0xffff);
+ /* workaround for broken BIOSes */
+ if (apm_info.bios.cseg_len <= apm_info.bios.offset)
+ _set_limit((char *)&cpu_gdt_table[i][APM_CS >> 3], 64 * 1024 -1);
+ if (apm_info.bios.dseg_len <= 0x40) { /* 0x40 * 4kB == 64kB */
+ /* for the BIOS that assumes granularity = 1 */
+ cpu_gdt_table[i][APM_DS >> 3].b |= 0x800000;
+ printk(KERN_NOTICE "apm: we set the granularity of dseg.\n");
+ }
}
#endif
}


--
Hiroshi Miura --- http://www.da-cha.org/
NTTDATA Corp. Marketing & Business Strategy Planning Dept. --- [email protected]
Key fingerprint = 9117 9407 5684 FBF1 4063 15B4 401D D077 04AB 8617
-- My hacking life is happy as the day is long

2002-10-26 10:33:17

by Pavel Machek

Subject: Re: APM work around for bad bios.

Hi!

> I use CASIO CASSIOPEIA FIVA 101 and 103. (http://www.da-cha.org/fiva/fiva.html)
> this use Cyrix MediaGX and Award BIOS.
> This APM BIOS report broken value for dseg_len.
> it asumes granularity is 1.
>
> I made a work around for this situation.
>
> * if (cseg_len < bios.offset) BIOS report BAD len value.
> -- segment length must be always larger than code offset
>
> * if (dseg_len <= 0x40 ) BIOS asumes granularity =1.
> -- 0x40 * 4kB = 64kB, my pc reports 0x40.
>
>
> diff -urB -x .config -x '*.[oasS]' -x '*.in' -x '*.rej' -x '*.orig' linux-2.5.43-orig/arch/i386/kernel/apm.c linux-2.5.43/arch/i386/kernel/apm.c
> --- linux-2.5.43-orig/arch/i386/kernel/apm.c 2002-10-12 13:21:05.000000000 +0900
> +++ linux-2.5.43/arch/i386/kernel/apm.c 2002-10-14 21:36:14.000000000 +0900
> @@ -1980,6 +2141,14 @@
> (apm_info.bios.cseg_16_len - 1) & 0xffff);
> _set_limit((char *)&cpu_gdt_table[i][APM_DS >> 3],
> (apm_info.bios.dseg_len - 1) & 0xffff);
> + /* workaround for broken BIOSes */
> + if (apm_info.bios.cseg_len <= apm_info.bios.offset)
> + _set_limit((char *)&cpu_gdt_table[i][APM_CS >> 3], 64 * 1024 -1);

Maybe add printk KERN_WARNING "apm: broken bios -- code segment too
short, assuming 64k"

> + if (apm_info.bios.dseg_len <= 0x40) { /* 0x40 * 4kB == 64kB */
> + /* for the BIOS that assumes granularity = 1 */
> + cpu_gdt_table[i][APM_DS >> 3].b |= 0x800000;
> + printk(KERN_NOTICE "apm: we set the
> granularity of dseg.\n");

Maybe better KERN_WARNING "apm: broken bios -- assuming granularity 1
on dseg"
Pavel
--
Worst form of spam? Adding advertisment signatures ala sourceforge.net.
What goes next? Inserting advertisment *into* email?