Regression in 2.4.20 radeonfb.c

In 2.4.19-rc3-ac2, Alan Cox / James Mayer added a modeline entry to modedb.c that enables a 1280x600 console screen on Vaio C1M* devices. The patch works fine for me in 2.4.19.

In 2.4.20-pre* kernels, it does no longer work for me, the console remains at the default setting.

I identified the hunk below to be the culprit. That check was removed in 2.4.20-* kernels. When I revert the hunk below, adding back the null check, my system again switches to full screen console mode.

I propose to revert the hunk for 2.4.20.
Tested with 2.4.20-rc1.

Thanks,
Kai

(Please CC me on replys)

diff -ru 2419/drivers/video/radeonfb.c 2420/drivers/video/radeonfb.c
@@ -877,14 +871,6 @@
        /* mem size is bits [28:0], mask off the rest */
        rinfo->video_ram = tmp & CONFIG_MEMSIZE_MASK;
 
-       /* According to XFree86 4.2.0, some production M6's return 0
-          for 8MB. */
-       if (rinfo->video_ram == 0 &&
-           (pdev->device == PCI_DEVICE_ID_RADEON_LY ||
-            pdev->device == PCI_DEVICE_ID_RADEON_LZ)) {
-           rinfo->video_ram = 8192 * 1024;
-         }
-
        /* ram type */
        tmp = INREG(MEM_SDRAM_MODE_REG);
        switch ((MEM_CFG_TYPE & tmp) >> 30) {

This "According to XFree86 4.2.0..." patch was never in the 2.4 tree. It is in the -ac tree (the latest -ac kernel is 2.4.20-pre10-ac2).

It might be a good idea to add it in 2.4.21-pre to the 2.4 tree but it's most likely too late to get it into 2.4.20.

cu
Adrian

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

For what it's worth, Ani Joshi was going to add it in the next radeonfb 2.4 version...

- James