On 21/01/2019 15:52, Wei Liu wrote: > On Mon, Jan 21, 2019 at 03:37:20PM +0000, Andrew Cooper wrote: >> panic() doesn't contain any caller information, so the sum output of: >> >> (d1) (XEN) >> (d1) (XEN) **************************************** >> (d1) (XEN) Panic on CPU 0: >> (d1) (XEN) Magic value is wrong: 336ec568 >> (d1) (XEN) **************************************** >> (d1) (XEN) >> >> isn't helpful at identifying what went wrong. Update the panic() strings to >> identify PVH and aid with diagnostics. >> >> The BUG_ON() check for ARRAY_SIZE(pvh_mbi_mods) is off-by-one, and redundant >> with the earlier panic() which explains things in more detail. Drop it. >> >> Finally, Xen takes nr_modules != 0 to mean that modlist_paddr is valid, but a >> cleverly crafterd PVH start info layout can cause Xen to use modlist_paddr at >> gaddr 0, in violation of the PVH spec. > Do you mean "using gaddr 0" is violation of spec? But I don't seem to > find it written down in the header file.
a gaddr of 0 indicates no modlist (per the general statement of addresses being zero), but Xen will use this if nr_modules != 0. ~Andrew _______________________________________________ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
