tree d64a9cc79e70ab20035d0b2718bc78cfca8fddea
parent 4564ee4b5063097bf747af3a3f8270a28adff820
author Pali Rohár <pali.rohar@gmail.com> 1790159732 +0200
committer LIU Hao <lh_mouse@126.com> 1790329972 +0800
gpgsig -----BEGIN SSH SIGNATURE-----
 U1NIU0lHAAAAAQAAADMAAAALc3NoLWVkMjU1MTkAAAAgpNkDDBuVPiz0dPd+zGoqB2ET19
 YkAU0wAtjm07QRvk4AAAADZ2l0AAAAAAAAAAZzaGE1MTIAAABTAAAAC3NzaC1lZDI1NTE5
 AAAAQE1oZPPnZCSUsphxvz3p+yJNmSH2QBv18RoDFVq2j7+mJQHOPxfJgFNCb5Nxr5R9jf
 +/3bzq5lUhjdVz7StpUg4=
 -----END SSH SIGNATURE-----

gendef: Do not set stdcall @0 for non-stdcall functions parsed from existing def files

If there is a def file "lib1.def" with line "func1" and gendef was called
on lib2.dll which contains func2 symbol defined as PE forward to lib1.func1
then gendef produces "lib2.def" file with line "func2@0 = lib1.func1" even
when gendef was not called with -a option.

This behavior is wrong because func1 in lib1.def is not defined as stdcall
therefore gendef should not add stdcall decoration for func2.

This change fixes this issue, changes the way how gendef reads and parse
input def files.

Signed-off-by: LIU Hao <lh_mouse@126.com>