On Fri, 2013-02-08 at 15:33 +0000, David Woodhouse wrote:
On Fri, 2013-02-08 at 15:27 +0000, Ian Campbell wrote:
On Fri, 2013-02-08 at 14:50 +0000, Ian Campbell wrote:
even with that and CONFIG_DEBUG_LEVEL=2 I still don't see the expected "Using pmtimer" message.
pci_setup() doesn't run on coreboot or Xen (because they do PCI setup themselves first), and AFAICT this is the path which leads to the call to pmtimer_setup(), so I don't think pmtimer will be initialised on either Xen or coreboot.
Except that my patch at http://git.infradead.org/users/dwmw2/seabios.git/commitdiff/56014611 adds a call to find_pmtimer() right after copying the Xen tables.
So the find_pmtimer() function which was added by my other patch at http://git.infradead.org/users/dwmw2/seabios.git/commitdiff/5ba80fac ought to go grubbing around in the Xen-provided ACPI table, find the pmtimer, and set SeaBIOS up to use it.
From then on, everything timer-related in the BIOS ought to use the pmtimer instead of the TSC (with lots of corresponding debug output).
Setting debug level 4 would show some output from find_fadt() and find_pmtimer() when it's trawling the ACPI tables and looking for the pmtimer. You *do* put a pm_tmr_blk field into your FADT table, don't you? If not, then I suppose it's not surprising that SeaBIOS doesn't find it :)