Re: Kernel Options
T <[email protected]> Tue, 26 Nov 2019 15:22:20 -0700
| Newsgroups | gmane.os.netbsd.ports.prep |
|---|---|
| Message-ID | <[email protected]> |
On 11/19/19 6:57 PM, T wrote: > Hello, > I've been trying out some kernel options and wanted to document a few > things on here, and to perhaps see if other similar peer machines are > also having the same results. Here's a list of the options with brief > notes: > > options DIAGNOSTIC > #Causes the kernel to panic, after the bootloader stage. This was > enabled for the secondary kernel, did not have it enabled in the > bootloader kernel. > > options KGDB > #This will lead to compile errors if enabled. I was able to track down > the errors to some changes that were made that did not take OEA type > machines into consideration. I was able to move some IFDEF tags to > fix, although there was one error I could not figure out, but was able > to bypass in order to get the kernel to compile. I'll follow up with > more information/patch when I return to this issue. > > options BUFQ_READPRIO > #When this was enabled, it fixed a problem where the UI would appear > to become stuck. Large amounts of IO to external storage devices > through PCI bus adapters (i.e. SATA, USB) would trigger this problem. > I have updated this ticket with more specific information: > http://gnats.netbsd.org/33291 > > options BUFQ_PRIOCSCAN > #This causes the kernel to panic. I tried to enable it because many of > the other ports have it enabled by default and was curious if it would > also fix the problem mentioned directly above. I was told that this is > the better buffer queue strategy, which would explain why the above > buffer queue strategy is rarely enabled in the default kernel > configuration files of other ports. I have yet to try FCFS, but > probably will in the near future. Some background on " Buffer Queue > Strategies": > https://mail-index.netbsd.org/netbsd-users/2015/12/02/msg017432.html > > I will submit these to gnats if I can get some sort of confirmation > that these are reproducible on a different machine. If more > information is needed, let me know. I just noticed in the current source that BUFQ_PRIOCSCAN has been enabled by default in port-prep (others too), has anyone had issues with current on their machine? http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/arch/prep/conf/GENERIC?only_with_tag=MAIN As noted above, this option caused a kernel panic when it was enabled.