mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* Debugging (?) output in 2.4.14 breaks radeon framebuffer
@ 2001-11-18 15:32 Ingo Saitz
  0 siblings, 0 replies; only message in thread
From: Ingo Saitz @ 2001-11-18 15:32 UTC (permalink / raw)
  To: Linux Kernel Mailing List

[-- Attachment #1: Type: text/plain, Size: 1371 bytes --]

MoiN

On bootup when the kernel activates the framebuffer console all
its revious messages remain on the screen, the tux logo is
displayed in the top corner and all remaining boot messages are
printed in one line over the belly of the pinguin. After bootup
all other consoles are working correctly and I can clean up the
mess on the first console by logging in and running "reset".

The last line displayed before linux activates the framebuffer is

radeonfb: ref_clk=2700, ref_div=60, xclk=16000 from BIOS

The next for lines, which are printk()d in radeon_write_mode():

CRTC_H_TOTAL_DISP = 0x4f0063, H_SYNC = 0x8c02a2
CRTC_V_TOTAL_DISP = 0x1df020c, V_SYNC = 0x8201ea
PPLL_DIV_3 = 0x301bf, PPLL_REF_DIV = 0x3c
DDA_CONFIG = 0x108032f, DDA_ON_OFF = 0x3e85924

seem to be the offending ones, because they are printed on the
screen _during_ the setup of the registers. Commenting them out
(see attached patch) does revert to a working console in
framebuffer mode.

My system: I have a Radeon (QD) SDR 32MB on AGP and a PII Celeron
533MHz on Intel LX Board (Elitegroup P6LX-A+)

    Ingo
-- 
/* So there I am, in the middle of my `netfilter-is-wonderful' talk in
 * Sydney, and someone asks `What happens if you try to enlarge a 64k
 * packet here?'.  I think I said something eloquent like `fuck'.  */
               -- net/ipv4/netfilter/ip_nat_ftp.c in linux 2.3.99-pre2

[-- Attachment #2: radeon-bf.diff --]
[-- Type: text/plain, Size: 533 bytes --]

diff -Ndurp linux/drivers/video/radeonfb.c linux/drivers/video/radeonfb.c
--- linux/drivers/video/radeonfb.c	Sun Nov 18 13:24:46 2001
+++ linux/drivers/video/radeonfb.c	Sun Nov 18 13:20:10 2001
@@ -2347,7 +2347,7 @@ static void radeon_write_mode (struct ra
 	OUTREG(CRTC_OFFSET, 0);
 	OUTREG(CRTC_OFFSET_CNTL, 0);
 	OUTREG(CRTC_PITCH, mode->crtc_pitch);
-#if 1
+#if 0
 	printk("CRTC_H_TOTAL_DISP = 0x%x, H_SYNC = 0x%x\n",
 		mode->crtc_h_total_disp, mode->crtc_h_sync_strt_wid);
 	printk("CRTC_V_TOTAL_DISP = 0x%x, V_SYNC = 0x%x\n",

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2001-11-18 23:49 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2001-11-18 15:32 Debugging (?) output in 2.4.14 breaks radeon framebuffer Ingo Saitz

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®