From mboxrd@z Thu Jan 1 00:00:00 1970 From: Bruno Ducrot Subject: Re: [PATCH][powernow-k7] correcting SGTC. Timer is based upon FSB. Date: Wed, 14 Apr 2004 14:47:17 +0200 Sender: cpufreq-bounces@www.linux.org.uk Message-ID: <20040414124717.GD8953@poupinou.org> References: <20040414103646.GB8953@poupinou.org> <20040414104148.GC24970@redhat.com> <20040414121857.GC8953@poupinou.org> <20040414122321.GH24970@redhat.com> Mime-Version: 1.0 Return-path: Content-Disposition: inline In-Reply-To: <20040414122321.GH24970@redhat.com> List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: cpufreq-bounces+glkc-cpufreq=gmane.org@www.linux.org.uk Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Dave Jones Cc: cpufreq@www.linux.org.uk On Wed, Apr 14, 2004 at 01:23:22PM +0100, Dave Jones wrote: > On Wed, Apr 14, 2004 at 02:18:57PM +0200, Bruno Ducrot wrote: > > Already? Thanks! > > I actually wanted to submit also the ACPI integration... > > Ok, I'll look at folding that in too after lunch. > > > Warning: it will only half work on the ACER Aspire, though: there is > > no pstate in the DSDT corresponding to the max frequency... > > I'm looking how to handle that correctly, probably at the init stage, > > if the max frequency is not given, then add it to powernow_table. It's > > really ugly, but I don't know how to make that OK. > > Suggestion is of course welcome.. > > Blacklist its BIOS with DMI tables, and adjust accordingly? Yep. Patch below (or I can send you the dmi output privately if you prefer). The machine is an ACER Aspire, look at: http://bugzilla.kernel.org/show_bug.cgi?id=2025 > Is it fixed maybe in a BIOS update ? I don't think so. Advice already done, don't know if the owner have done that though. > > > * Integrate acpi perflib from Dominik. > > * Use acpi if the PST tables are known to be broken (ASUS Aspire match one PST, > > but give brain damaged values), or use ACPI if no PST found. > > Was that the one that had every entry exactly the same ? It give: powernow: cpuid: 0x781 fsb: 133 maxFID: 0x1 startvid: 0xb powernow: FID: 0x4 (5.0x [665MHz]) VID: 0xc (1.400V) powernow: FID: 0x6 (6.0x [798MHz]) VID: 0xc (1.400V) powernow: FID: 0x8 (7.0x [931MHz]) VID: 0xc (1.400V) powernow: FID: 0x9 (7.5x [997MHz]) VID: 0xc (1.400V) powernow: FID: 0x9 (7.5x [997MHz]) VID: 0xc (1.400V) Whereas actually it's a mobile (startvid is 0xb. So, voltage can be scaled). And there is 'only' two entries with same values. With ACPI, (not tested yet) it should give that: FID: 0xd (9.5x [1263.500MHz] VID: 0xc (1.400mV) FID: 0xa (8.0x [1064.000MHz] VID: 0xd (1.350mV) FID: 0x8 (7.0x [931.000MHz] VID: 0xe (1.300mV) FID: 0x6 (6.0x [798.000MHz] VID: 0x11 (1.250mV) FID: 0x4 (5.0x [665.000MHz] VID: 0x13 (1.200mV) It's better, but max still not here. --- linux-dj-cpufreq/arch/i386/kernel/dmi_scan.c 2004/04/14 12:28:28 1.1 +++ linux-dj-cpufreq/arch/i386/kernel/dmi_scan.c 2004/04/14 12:29:42 @@ -921,6 +921,12 @@ static __initdata struct dmi_blacklist d NO_MATCH, NO_MATCH, } }, + { acer_cpufreq_pst, "Acer Aspire", { + MATCH(DMI_SYS_VENDOR, "Insyde Software"), + MATCH(DMI_BIOS_VERSION, "3A23"), + NO_MATCH, NO_MATCH, + } }, + #ifdef CONFIG_ACPI_BOOT /* * If your system is blacklisted here, but you find that acpi=force -- Bruno Ducrot -- Which is worse: ignorance or apathy? -- Don't know. Don't care.