Date: Wed, 29 Oct 2008 19:40:39 +0200 From: Serge Semenenko <serge@a-1.com.ua> To: John Baldwin <jhb@freebsd.org> Cc: freebsd-acpi@freebsd.org, Alexander Motin <mav@freebsd.org> Subject: Re: PCIe bridges resources disappearing with ACPI enabled. Message-ID: <4908A017.3070305@a-1.com.ua> In-Reply-To: <200810281108.56218.jhb@freebsd.org> References: <4903A120.7040003@FreeBSD.org> <200810271142.59666.jhb@freebsd.org> <4906450E.4030700@a-1.com.ua> <200810281108.56218.jhb@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
John Baldwin wrote: > On Monday 27 October 2008 06:47:42 pm Serge Semenenko wrote: > >> John Baldwin wrote: >> >>> On Saturday 25 October 2008 06:43:44 pm Alexander Motin wrote: >>> >>> >>>> Hi. >>>> >>>> I have spent whole day trying to investigate strange problem of my Acer >>>> TM6292 laptop (965GM+ICH8M). When booted with ACPI enabled, all three >>>> of PCIe-to-PCIe bridges appearing completely without I/O resources: >>>> pcib1: <ACPI PCI-PCI bridge> irq 17 at device 28.0 on pci0 >>>> pcib1: domain 0 >>>> pcib1: secondary bus 2 >>>> pcib1: subordinate bus 3 >>>> pcib1: I/O decode 0x0-0x0 >>>> pcib1: no prefetched decode >>>> ... >>>> At the same time, with ACPI disabled, resources are present. There are >>>> some different problem with IRQ in that case, but it is another >>>> question, not so interesting to me. >>>> >>>> I have tried both IO and memory mapped PCIe configuration registers >>>> without success. >>>> >>>> I have made heavy digging trying to find where resources disappearing. I >>>> have even added debug printing inside pcireg_cfgwrite() and >>>> pciereg_cfgwrite() to trace if somebody erases it and found nothing. >>>> Nothing writes into that devices configuration registers. >>>> >>>> >>> The SMI handle could be clearing the BARs when ACPI is enabled for some >>> reason. Windows and Linux are smart enough to alloc resources for >>> > bridges, > >>> but FreeBSD isn't yet. >>> >>> >>> >> Thanks for a good tip. Elimination of "Store (Zero, SMIC)" string from >> ASL code has solved the problem. >> > > That may not be a good solution though as there may be other things the SMI > handler is doing to enable ACPI support that the OS is dependent on. > > Seems you're right. Some WMI routines stop working. But anyway that appears unrelated to the subject. Got to make a confession... It's a shame, but I've forgot about NO_MODULES="YES" option at make.conf... So my acpi.ko remains hacked all the time... After rebuild even with patched ASL bridge resources are still unassigned... Sorry for the mess..
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4908A017.3070305>