Bug 7866
| Summary: | cpufreq fails to correctly identify Intel Conroe operating frequencies | ||
|---|---|---|---|
| Product: | Power Management | Reporter: | Jared Luxenberg (jared) |
| Component: | cpufreq | Assignee: | cpufreq (cpufreq) |
| Status: | REJECTED WILL_NOT_FIX | ||
| Severity: | high | CC: | acpi-bugzilla |
| Priority: | P2 | ||
| Hardware: | i386 | ||
| OS: | Linux | ||
| Kernel Version: | 2.6.20 | Subsystem: | |
| Regression: | --- | Bisected commit-id: | |
| Attachments: | result of acpidump | ||
|
Description
Jared Luxenberg
2007-01-23 00:06:04 UTC
Please post acpidump output Created attachment 10167 [details]
result of acpidump
Here is the output of acpidump
I've updated to 2.6.20 final and I'm still having the same problem. Jared, What frequency is set for your FSB? is it 266/1066 or something else? Some people reported that they see such frequency values from BIOS if they (or mainboard manufacturer) overclocked the FSB. Could you check if you see right values if the FSB is set to default above? The FSB is currently overclocked, I think I have it set at 350 or so. However, I was still having this problem before I overclocked it, when the FSB was at 266. I can set the FSB back to 266 and test that again with 2.6.20 if you think that's the problem (I suspect it's not). But cpufreq should probably either disable itself or work correctly because the failure mode is pretty bad. I wouldn't have noticed the problem if I hadn't cat-ed /proc/cpuinfo. The FSB is currently overclocked, I think I have it set at 350 or so. However, I was still having this problem before I overclocked it, when the FSB was at 266. I can set the FSB back to 266 and test that again with 2.6.20 if you think that's the problem (I suspect it's not). But cpufreq should probably either disable itself or work correctly because the failure mode is pretty bad. I wouldn't have noticed the problem if I hadn't cat-ed /proc/cpuinfo. cpufreq works. it switches between right frequencies (they are programmed as multipliers for FSB freq). The only "broken" thing is the output of cpufreq, there it prints these (multipliers * lowest FSB freq, which is 100Mhz). Values to print cpufreq takes from BIOS, so it is BIOS who used 100Mhz instead of your 350 for above calculation. 6-9 is a valid range for E6600. if you multiply it by default 266 you will get advertised frequencies of E6600, if you multiply them by 350, you will have your real frequencies. cpufreq has no access to FSB frequency, so there is no option but to either live with such printouts, or bug your BIOS vendor about it. |