Re: benh 2.4.6 kernel

[gtgj@pacbell.net (Gregorio Gervasio Jr.)]

>>>>> On Tue, 10 Jul 2001 23:30:48 -0700, gtgj@pacbell.net (Gregorio Gervasio Jr.) said:

[ TiPB sleep problems with 2.4.6 ]

g>         I get the same trace.  As Olaf Hering suggested, it's failing
g> when "giveup_altivec" is called.  It looks like a processor version
g> check was recently added to "giveup_altivec" (in
g> arch/ppc/kernel/head.S) but it seems to be incorrect:

g> 	.globl	giveup_altivec
g> giveup_altivec:
g> 	mfpvr	r24			/* check if we are on a G4 */
g> 	srwi	r24,r24,16
g> 	cmpwi	r24,PVR_7400@h
g> 	beq	3f			/* continue */
g> 	mflr	r24
g> 	bl	msr_vec_debug		/* debug thingy in process.c */
g> 	mtlr	r24
g> 	blr
g> 3:
g> ...

g> I think PVR_7400 is the unshifted value?  (Also, the TiPB uses a 7410,
g> so what is the right comparison?)  Anyway, without this check, sleep
g> seems to be fine in both console and X11 mode.

        A little more info:

        When I removed the debug code, console sleep started
failing again (as it does even with older kernels).  X11 sleep still
works.  It seems like the problem is in aty128_sleep_notifier because
this debug code you suggested:

> - edit drivers/video/aty128fb.c, comment out the call to
>   pmu_register_sleep_notifier() line 1907, and add
>   a "return;" at the beginning of aty128fbcon_blank() (line 2279)
>   so that the function does nothing.

allows console sleep to work (not completely of course because the
screen doesn't shut off).

Gregorio Gervasio, Jr.
gtgj@pacbell.net