fa78ecacec
The SCI handler for the GPE associated with the Super I/O did not clear the respective PME status bits resulting in the SCI reoccurring endlessly. The /proc/interrupts reported millions of ACPI interrupts generated in just a few minutes of uptime. The flood of interrupts caused some units to be unusable in extreme cases once attempted to boot Qubes OS for example. On systems like Qubes OS it had a huge impact on performance due to many IPCs the SCIs caused under Xen. Clear the PME bits of devices that report a PME event. Then clear the global PME status bit at the end of SCI handler to prevent the SCI from asserting again until a new event occurrs. With this change the number of ACPI interrupts generated in the first minutes of uptime settles at a few thousands. TEST=Boot Qubes OS R4.1.2 on Dell OptiPlex 9010 SFF and check /proc/interrupts in dom0 if the number of ACPI interrupts is only a few thousands. Change-Id: I64e03d268138a62b46084be41343ef7fb089dfc3 Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com> Reviewed-on: https://review.coreboot.org/c/coreboot/+/78351 Tested-by: build bot (Jenkins) <no-reply@coreboot.org> Reviewed-by: Krystian Hebel <krystian.hebel@3mdeb.com> |
||
|---|---|---|
| .. | ||
| acpi | ||
| aspeed | ||
| common | ||
| fintek | ||
| ite | ||
| nsc | ||
| nuvoton | ||
| renesas | ||
| smsc | ||
| winbond | ||