On Wed, 17 Oct 2007, Ralf Baechle wrote: > The two things are a know lose end. There is a bug in some old MIPS > processors where reading one of the compare or count registers in exactly > the moment when both have identical values in the interrupt getting lost. > > Will have to dig up the details on that one again before I can implement > a proper workaround ... This is the erratum #53 of the R4000PC/SC processor and it triggers if the Count register is read. Conveniently, in the errata sheet as distributed, the text is covered by a figure (Figure 1a on page 13), so you can only reach the page with some PostScript magic. I did it a while ago and now have a separate document available which provides the text of page 13 with the figure removed. I can provide it if there is interest. Maciej
