nexoniarz/SchOS
nexoniarz/SchOS
Watch
1
0
Fork
You've already forked SchOS
 0
Code Issues Pull requests Projects Releases 1 Packages Wiki Activity Actions

acpi, vga: say what the first real boot actually found

Browse source 
"acpi: 0 PCI interrupts routed through _PRT" on an AMD board meant that
every PCIe bridge's devices are wired to the second I/O APIC, which is not
driven. It says so now, and that those devices need MSI.

"vga: framebuffer disagreement" on an NVIDIA card compared the GOP's
framebuffer against BAR0, a 16 MB register block big enough to pass for a
screen. The BAR that contains the framebuffer is preferred now.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01S1giNvrHTrPMPw7V6x1kQ4
This commit is contained in:
Nexoniarz 2026-10-22 23:47:17 +02:00
commit 416aa52b09
2 changed files with 20 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

10
kernel/acpi/aml.c
View file

@ -284,7 +284,7 @@ void
acpi_pci_route(void)
{
uacpi_pci_routing_table *roottab;
int i, routed = 0;
int i, routed = 0, beyond = 0;
if (!ready)
return;
@ -316,8 +316,11 @@ acpi_pci_route(void)
if (gsi < 0)
continue;
if (gsi >= NIRQ_IOAPIC) {
// A second I/O APIC's input. Not reachable here; MSI still is.
// A second I/O APIC's input. Not reachable here; MSI still is. AMD
// boards wire every PCIe bridge's devices to the GNB's I/O APIC, so
// on those this is most of the machine.
d->irq = 0xFF;
beyond++;
continue;
}
if (gsi != d->irq)
@ -331,4 +334,7 @@ acpi_pci_route(void)
uacpi_free_pci_routing_table(roottab);
kinfo("acpi: %d PCI interrupt%s routed through _PRT\n", routed,
routed == 1 ? "" : "s");
if (beyond)
kinfo("acpi: %d more on a second I/O APIC, which is not driven; "
"those devices need MSI\n", beyond);
}