Most recent kernel where this bug did *NOT* occur: 2.4.19 Distribution: Red Hat Enterprise Linux 4, Fedora Core 6, Ubuntu 5. Hardware Environment: IBM ThinkPad T23. See complete list of hardware at http://www-307.ibm.com/pc/support/site.wss/document.do?lndocid=MIGR-4YTG43 Software Environment: None. The bug occures right after the kernel have been uncompressed. Problem Description: A high frequent tone comes from the hardware near the USB connectors right after the 2.6.x kernel have been uncompressed. The tone is there everytime the harddrive loads or saves. When the harddrive doesn't do any of this, the tone is gone. Inserting a USB key inverses the problem. When a USB key is inserted, the tone is present when the harddisk doesn't load or save, and comes back when it loads and saves. I have tested it with two different harddrives, and it difficult to tell from what peice of hardware the tone comes from, but it is not from the harddrive. It is from somewhere near the USB connectors. Using the FSF membership cd LNX-BBC 2.1 with kernel 2.4.19-xfs i686 works perfectly. Steps to reproduce: * Boot any 2.6 kernel * Wait till the kernel is uncompressed, and a high frequent tone appears from the hardware.
This looks like a hardware problem, the kernel can't make "high frequent tones" Or do you mean the computer's speaker?
> This looks like a hardware problem, the kernel can't make "high frequent > tones" No, but the 2.6 kernel trickers some component to vibrate/generate a high frequcency tone. It doesn't happen with 2.4 kernels. > Or do you mean the computer's speaker? No, it is a coil or some other component that makes the noise that is not support to be able to make a sound.
Apparently you've hit the "high pitched noise" problem: http://lkml.org/lkml/2004/10/7/69 This is a hardware problem caused by a kernel 'feature'. HZ is the frequency of the timer interrupt; Linux 2.4 had HZ=100; 2.6 defaults to HZ=1000. Windows XP defaults to 100; which means that most of hardware works fine at HZ=100, but makes strange noises - because of cheap capacitators in the hardware - when moved to 1000. 2.4 works fine, 2.6 makes noises. There's a "timer frequency" kernel configuration option in the "processor type and features" section of the kernel that allows to set it to 1000. You probably need to recompile your own kernel . Or first try one of the "server" kernels available optionally in some distros, those server kernels default to HZ=100 because it can improve performance a bit.
"here's a "timer frequency" kernel configuration option in the "processor type and features" section of the kernel that allows to set it to 1000" I meant 100, of course!
Okay, this is very interesting! Can you give the path in makeconfig where I can find a setting for this? If there is no option for changing it, do you know which file and what I should change?
Sorry, I didn't read your last paragraf, which answers my last question =) Thanks for the fast replys =)
Just for confirm that it worked. Just compiled 2.6.20 where CONFIG_HZ=100 and there are no more high pitch tone.