Re: (tiny) patch review: return if RTC diag fails
Manuel Bouyer <[email protected]>
| Newsgroups | gmane.os.netbsd.devel.kernel |
|---|---|
| Message-ID | <[email protected]> |
On Thu, Feb 13, 2025 at 06:29:03PM +0100, Emile `iMil' Heitor wrote: > On Thu, 13 Feb 2025, Manuel Bouyer wrote: > > > I think it's not right: mc146818_read(NULL, NVRAM_DIAG) returning a non--0 > > value doens't means that is fails. Usually the RTC BIOS diagnostic has some > > error because the chip lost track of date/time (maybe because of a bad > > battery). In this case you still want to use the RTC. > > Understood. So as this seems to only impact qemu/microvm should we > consider an if (vm) test? > > > Did you check what value is returned in your case ? > > 255 Looks like there's nothing to be read here; some of the bits should never be 1 (see NVRAM_DIAG_BITS definition in nvram.h). This is what inb returns when there's nothing at this address. so maybe test if (vm && s == 0xff) ? -- Manuel Bouyer <[email protected]> NetBSD: 26 ans d'experience feront toujours la difference --