On 27.11.2025 20:31, Andrew Cooper wrote: > On 26/11/2025 1:48 pm, Jan Beulich wrote: >> --- a/xen/scripts/Kbuild.include >> +++ b/xen/scripts/Kbuild.include >> @@ -56,6 +56,19 @@ define filechk >> fi >> endef >> >> +### >> +# Compare the symbol tables of two object files. As diff's -I option isn't >> +# standardized, the name difference of the two object files needs >> abstracting >> +# out. >> +define compare-symbol-tables >> + ln -f $(1) $(@D)/.cst.$$$$; \ >> + $(OBJDUMP) -t $(@D)/.cst.$$$$ > $(1).sym; \ >> + ln -f $(1) $(@D)/.cst.$$$$; \ >> + $(OBJDUMP) -t $(@D)/.cst.$$$$ > $(2).sym; \ >> + rm -f $(@D)/.cst.$$$$ > > Doesn't the second link want to be `ln -f $(2)` ? > > It looks like this is comparing $(1) to itself.
Ouch. Now I'll be curious if with this corrected I still see this pass successfully. >> + diff -u $(1).sym $(2).sym >> +endef > > While I am generally in favour of build time checks, this looks like > it's a large overhead? I don't think it's really noticeable in the overall link time, considering how slow especially the xen.efi linking is. Jan
