On 27/02/2017 06:53, Alexey Kardashevskiy wrote: > There is a link to this one [1]: > http://git.qemu-project.org/?p=qemu.git;a=commitdiff;h=c3c1bb99d1c11978d9ce94d1bdcf0705378c1459 > > Which was reverted with a sensible explanation [2]: > http://git.qemu-project.org/?p=qemu.git;a=commitdiff;h=4025446f0ac6213335c22ec43f3c3d8362ce7286 > > However it is still in the tree as [3]: > http://git.qemu-project.org/?p=qemu.git;a=commitdiff;h=23820dbfc79d1c9dce090b4c555994f2bb6a69b3 > > The only difference between [1] and [3] is a fixed typo in the subject, > other than that they are identical. > > Is not the explanation from [2] correct any more and [3] is a correct final > fix? Or [3] should not be in the tree at all?
The explanation from [2] is correct, it caused "problems with boards that declare memory regions shorter than the registers they contain" and those were fixed. Unfortunately I remember that the fixes were complex, IIRC covering mostly ioport.c and Xen. It might be fine for pSeries though. Paolo
