In OpenSuSE 10.2 after installing ndiswrapper, the system would always hang during bootup or shutdown at hwclock. I reverted to the 32 bit version of hwclock and the system was stable. I put an entry in bugzilla.novell.com. HWclock behaves differently in 64 bit mode vs 32 bit.
I have been working with 10.3 Alpha and due to the fact that bcm43xx is not ready for primetime, I use ndiswrapper (1.45) for testing. If the system is configured to use ndiswrapper, it is loaded before hwclock runs, and the system will lock tight, as hwclock gets lost in a loop waiting for a interrupt to occur. If I let bcm43xx load then rmmod it and the ieee80211 modules, then load ndiswrapper, wireless works.
The question is what effect, if known, does ndiswrapper have on clock functions that would cause the observed issues with hwclock?
This is the bug report I filled
That has better information on the differences in hwclocks 32/64 operations.
The 64 bit version polls while the 32 bit version looks at an interrupt.
If ndiswrapper loads then hwclock(64) polling goes into a loop that it never exits.
If ndiswrapper does not load before hwclock runs, then it succeeds.
Your description doesn't give clues to what the problem could be. How do you know 'hwclock gets lost in a loop waiting for interrupt'? Which hwclock? User space application? I don't see why that should wait for an interrupt. And it shouldn't affect any kernel driver.
Log in to post a comment.
Sign up for the SourceForge newsletter:
You seem to have CSS turned off.
Please don't fill out this field.