From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jeffrey A Law To: Mark Klein Cc: gcc@gcc.gnu.org Subject: Re: MPE Port Date: Thu, 14 Oct 1999 04:03:00 -0000 Message-id: <30257.939898754@upchuck.cygnus.com> References: <4.2.0.58.19991009200725.00c5a100@garfield.dis.com> X-SW-Source: 1999-10/msg00287.html In message < 4.2.0.58.19991009200725.00c5a100@garfield.dis.com >you write: > MPE doesn't use a crt.o or anything like that where I could "hide" this, so > I'm faced with explicitly using a separate .o file for this purpose. Seems > to me it was easier to use the same technique as the reference to > __gcc_plt_call > in the DO_GLOBAL_DTORS_BODY macro than trying to jump through all these hoo > ps > to make this a global constructor. No, it is *totally* inappropriate for this behavior to be forced by GCC. This has to be done elsewhere. If that means explicitly adding a .o to the link line to get this behavior, well, then that's what'll have to happen. I can't stress enough that it is totally wrong for us to try and do this within GCC. jeff From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jeffrey A Law To: Mark Klein Cc: gcc@gcc.gnu.org Subject: Re: MPE Port Date: Sun, 31 Oct 1999 23:35:00 -0000 Message-ID: <30257.939898754@upchuck.cygnus.com> References: <4.2.0.58.19991009200725.00c5a100@garfield.dis.com> X-SW-Source: 1999-10n/msg00287.html Message-ID: <19991031233500.R5WE18BnkXxFWX1QZLRsrFJfdOW9orVyqbV6YlxwORQ@z> In message < 4.2.0.58.19991009200725.00c5a100@garfield.dis.com >you write: > MPE doesn't use a crt.o or anything like that where I could "hide" this, so > I'm faced with explicitly using a separate .o file for this purpose. Seems > to me it was easier to use the same technique as the reference to > __gcc_plt_call > in the DO_GLOBAL_DTORS_BODY macro than trying to jump through all these hoo > ps > to make this a global constructor. No, it is *totally* inappropriate for this behavior to be forced by GCC. This has to be done elsewhere. If that means explicitly adding a .o to the link line to get this behavior, well, then that's what'll have to happen. I can't stress enough that it is totally wrong for us to try and do this within GCC. jeff