Hi, Vivek Sorry for my late response. On 03/20/14 at 11:42am, Vivek Goyal wrote: > On Wed, Mar 19, 2014 at 04:03:57PM +0800, WANG Chao wrote: > > Hi, All > > > > When kaslr comes in and kdump is broken, it seems about the right time to use > > E820 instead of memmap=exactmap to pass memmap for kdump for the default memmap > > passing mechanism: > > http://lists.infradead.org/pipermail/kexec/2014-February/011048.html > > What's the connection between kaslr and passing memory mappings in > bootparams? kaslr doesn't work with memmap=exactmap. kaslr happens before parsing user defined memmap and it walks through E820 in boot_params to determine 2nd kernel text area. And it would override the 1st kernel's ram. > > To me this is completely independent of kaslr. This is moving towards > a better design to pass memory map to second kernel which in turn will > also solve the issue of too many memory ranges on large machines. Yes. You're right. That's what this patchset is mostly for. kASLR is just one of the reasons. > > Also, could you get hold of a machine with calgary IOMMU and test that > --pass-memmap option makes those calgary machines work? I couldn't find any calgary box. I tried to force calgary iommu on some IBM-X series, but didn't work out. Those old hardwares are really hard to find these days. I think given the fact that the calgary code chage is simple enough and hpa has committed to tip tree, this patchset won't risk much. Thanks WANG Chao
