pc/sis630: more BIOS notes

This commit is contained in:
angelosa 2024-03-23 02:26:06 +01:00
parent b77af1eb38
commit f338e1590e

View File

@ -3,57 +3,61 @@
// thanks-to: Diego Nappino
/**************************************************************************************************
SiS 630 chipset based PC
SiS 630 chipset based PC
TODO (main):
- Finalize Super I/O handling of ITE 8705F (I/O $2e / $2f)
\- fan monitor, cfr. I/O $294 reads in shutms11 BIOS fan tests;
\- FDC doesn't work, moans on first boot;
- '900 Ethernet PXE (missing ROM dump);
- USB controllers (OpenHCI complaint);
- ACPI is not fully lpc-acpi complaint;
- EISA slots;
- SMBus;
- PS/2 mouse is unstable, worked around by disabling and using a serial mouse instead.
TODO (main):
- Finalize Super I/O handling of ITE 8705F (I/O $2e / $2f)
\- fan monitor, cfr. I/O $294 reads in shutms11 BIOS fan tests;
\- FDC doesn't work, moans on first boot;
- '900 Ethernet PXE (missing ROM dump);
- USB controllers (OpenHCI complaint);
- ACPI is not fully lpc-acpi complaint;
- EISA slots;
- SMBus;
- PS/2 mouse is unstable, worked around by disabling and using a serial mouse instead.
TODO (usability, to be moved in a SW list):
- windows xp sp3: tests HW then does an ACPI devtrap write ($48), will eventually BSoD with
ACPI STOP #a5 error with param $11
\- To bypass hold F7 while the "to install SCSI drivers [...] press F6" appears.
And by F7 I really mean it :shrug:
TODO (usability, to be moved in a SW list):
- windows xp sp3: tests HW then does an ACPI devtrap write ($48), will eventually BSoD with
ACPI STOP #a5 error with param $11
\- To bypass hold F7 while the "to install SCSI drivers [...] press F6" appears.
And by F7 I really mean it :shrug:
- windows xp sp3: BSoD during install with a STOP #0a IRQL_NOT_LESS_OR_EQUAL;
- windows xp sp3: BSoD during install with a STOP #0a IRQL_NOT_LESS_OR_EQUAL;
- windows neptune: BSoD during ethernet check (after time clock setup)
with a STOP #a0 INTERNAL_POWER_ERROR with param1 0x5 ("reserved"!?)
- windows neptune: BSoD during ethernet check (after time clock setup)
with a STOP #a0 INTERNAL_POWER_ERROR with param1 0x5 ("reserved"!?)
- gamecstl Kontron BIOS:
\- hangs at PC=0xf3cf2, again wanting a SMI# from devtrap_en_w;
\- No PS/2 inputs;
- gamecstl Kontron BIOS:
\- hangs at PC=0xf3cf2, again wanting a SMI# from devtrap_en_w;
\- No PS/2 inputs;
- gamecstl dump (tested from shutms11, also see notes below):
\- Currently black screens before booting normal Windows, reading $5004 from the LPC ACPI
(flip EAX to non-zero to bypass).
NB: it also writes to $5048 once (devtrap_en_w), which should generate a SMI# event;
\- Doesn't accept any PS/2 input, tries to install a "PCI standard CPU Host Bridge" (?),
hangs there;
\- GUI is never recognized no matter what, punts with DirectX not installed;
- gamecstl dump (tested from shutms11, also see notes below):
\- Currently black screens before booting normal Windows, reading $5004 from the LPC ACPI
(flip EAX to non-zero to bypass).
NB: it also writes to $5048 once (devtrap_en_w), which should generate a SMI# event;
\- Doesn't accept any PS/2 input, tries to install a "PCI standard CPU Host Bridge" (?),
hangs there;
\- GUI is never recognized no matter what, punts with DirectX not installed;
- xubuntu 6.10: throws several SCSIDEV unhandled $46 & $51 commands
(get configuration/read disc information),
eventually punts to prompt with a "can't access tty: job control turned off" (on live CD) or
hangs at "Configuring network interfaces" (on actual install);
- xubuntu 6.10: throws several SCSIDEV unhandled $46 & $51 commands
(get configuration/read disc information),
eventually punts to prompt with a "can't access tty: job control turned off" (on live CD) or
hangs at "Configuring network interfaces" (on actual install);
- xubuntu 10.10: stalls after '900 ethernet check;
- xubuntu 10.10: stalls after '900 ethernet check;
- Haiku 0.1: hangs throwing an "unhandled READ TOC format 2",
serial COM1 prints a "vm_mark_page_range_inuse: page 0x9f in non-free state 7!"
- Haiku 0.1: hangs throwing an "unhandled READ TOC format 2",
serial COM1 prints a "vm_mark_page_range_inuse: page 0x9f in non-free state 7!"
- Red Hat 6.2: Triple Faults on x87 exception check
\- prints the type of mounted floating point exception if bypassed.
- Red Hat 6.2: Triple Faults on x87 exception check
\- prints the type of mounted floating point exception if bypassed.
NOTES:
- BeOS 5 hardwires serial mouse checks at $2e8, ignoring BIOS PnP.
Notes on possible shutms11 BIOS bugs:
- BeOS 5 hardwires serial mouse checks at $2e8, ignoring BIOS PnP.
- FreeDOS 1.3 ctmouse often fails serial detection (will work if you load it at prompt)
- BIOS in general often throws serial conflicts when changing port setups,
even on perfectly valid combinations (i.e. COM1 assigned while COM2 is *disabled*)
- PCI device listing often forgets to list print ACPI irq.
===================================================================================================