RE: [Dri-devel] Linux kernel PCI IDs vs Xfree
From: Alexander Stohr <hidden>
Date: 2003-08-09 20:41:01
all info contained below is my personal opinion, should be retriveable from several pbulic sources and it is only accurate to my best knowledge.
linux/pci_ids.h and xf86PciInfo.h are in disagreement for several Rage128 PCI ids. Xfree appears to be correct and the kernel file is wrong. This impacts the Rage framebuffer driver when loading.
as long as no one notices or complains it cant be that big issue. ;-)
I verified these with two PCI databases.
Which ones? e.g. www.yourvote.com\pci is the database where the Linux kernel does get its listing. www.plasma-online.de is another site with a more or less bigger emphasize on getting even a photo of the packages.
Rage Mobility 128 AGP 4x #define PCI_CHIP_RAGE128MF 0x4D46 #define PCI_DEVICE_ID_ATI_RADEON_LF 0x4d46
this chipset possibly went trough press announcements as "Mobility M4, AGP 4x". i dont know the board names so the "Rage Mobility 128" might match. 0x4d46 is ascii equivalent of "MF", so the 2nd line sounds odd. Official Radeon-1 is counted (to my understanding) as equivalent to a Rage in 6th revison, so a 4 cant be a Radeon at all.
Rage Mobility 128 AGP #define PCI_CHIP_RAGE128ML 0x4D4C missing
"Mobility M4, AGP 4x"
Rage 128 SE/4x PCI #define PCI_CHIP_RAGE128SE 0x5345 #define PCI_DEVICE_ID_ATI_RAGE128_RM 0x5345
0x5345 == ASCII "SE" Rage 128 4x, PCI
Rage 128 SF/4x AGP 2x #define PCI_CHIP_RAGE128SF 0x5346 #define PCI_DEVICE_ID_ATI_RAGE128_RN 0x5346
0x5346 == ASCII "SF" Rage 128 4x, AGP 2x
Rage 128 SG/4x AGP 4x #define PCI_CHIP_RAGE128SG 0x5347 #define PCI_DEVICE_ID_ATI_RAGE128_RO 0x5347
0x5347 == ASCII "SG" Rage 128 4x, AGP 4x the 4x in the chip/card name is more of the evolution stage of the asic than the realized PCI/AGP bus speed. no idea how that RM trough RO came into the given macros.
Rage 128 SH/4x #define PCI_CHIP_RAGE128SH 0x5348 missing
0x5348 == ASCII "SH" dont know if there was ever build such a chip revison.
Rage 128 SK/4x #define PCI_CHIP_RAGE128SK 0x534B #define PCI_DEVICE_ID_ATI_RAGE128_RG 0x534b
0x534B == ASCII "SK" Rage 128 4x, PCI
Rage 128 SL/4x AGP 2x #define PCI_CHIP_RAGE128SL 0x534C #define PCI_DEVICE_ID_ATI_RAGE128_RH 0x534c
0x534C == ASCII "SL" Rage 128 4x, AGP 2x
Rage 128 SM/4x AGP 4x #define PCI_CHIP_RAGE128SM 0x534D #define PCI_DEVICE_ID_ATI_RAGE128_RI 0x534d
0x534D == ASCII "SM" Rage 128 4x, AGP 4x the 4x in the chip/card name is more of the evolution stage of the asic than the realized PCI/AGP bus speed. no idea how that RG trough RI came into the given macros.
Rage 128 4x #define PCI_CHIP_RAGE128SN 0x534E missing
0x534E == ASCII "SN" dont know if there was ever build such a chip revison.
Rage 128 Pro Ultra TF #define PCI_CHIP_RAGE128TF 0x5446 #define PCI_DEVICE_ID_ATI_RAGE128_U1 0x5446
0x5446 == ASCII "TF" AGP 4x. might have flat panel interface -> possibly laptop. Rage128 series, cant tell about board name. I cant tell if the "U1" is the correct chip name but if considered an integrated design then might fit.
Rage 128 Pro Ultra TL #define PCI_CHIP_RAGE128TL 0x544C #define PCI_DEVICE_ID_ATI_RAGE128_U2 0x544C
0x544C == ASCII "TL" AGP 4x. might have flat panel interface -> possibly laptop. Rage128 series, cant tell about board name. I cant tell if the "U2" is the correct chip name but if considered an integrated design then might fit.
Rage 128 Pro Ultra TS #define PCI_CHIP_RAGE128TS 0x5453 missing
0x5453 == ASCII "TS" Rage128 series, cant tell about board name. I doubt the Ultra - it could even be a used for a Rage Fury MAXX board codenamed Aurora (see toms hardware) and later announced as a dual "RAGE 128 PRO" design.
Rage 128 Pro Ultra TT #define PCI_CHIP_RAGE128TT 0x5454 missing
0x5454 == ASCII "TT" Rage128 series, cant tell about board name. I doubt the Ultra - it could even be a used for a Rage Fury MAXX board codenamed Aurora (see toms hardware) and later announced as a dual "RAGE 128 PRO" design.
Rage 128 Pro Ultra TU #define PCI_CHIP_RAGE128TU 0x5455 missing
0x5455 == ASCII "TU" Rage128 series, cant tell about board name. I doubt the Ultra - it could even be a used for a Rage Fury MAXX board codenamed Aurora (see toms hardware) and later announced as a dual "RAGE 128 PRO" design. as initially written, no one complained for ages, so there is quite minor interest from users and maybe not even hardware availabel to developer hands to verify those chipsets and boards in the real world of x86 computer systems anymore. answered in best way of what i could find about those historic designs and what the world wide web plus the known device id encoding rules do tell me and anyone. anyways thats written on the fly and can be badly false but in the hope to contribute to discussion in a helpfull way. Jon, what are you planning to do with that now? Will your final ratio become a patch against the XF86-CVS or the linux kernel sources? Will you contribute your result to e.g. yourvote.com? -Alex. ------------------------------------------------------- This SF.Net email sponsored by: Free pre-built ASP.NET sites including Data Reports, E-commerce, Portals, and Forums are available now. Download today and enter to win an XBOX or Visual Studio .NET. http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01