This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Bootstrap Failure on x86-Linux With "-pipe"

Sorry Andrew, after much digging I find that your
patch indeed is the right solution till such a time
that we decide what to do about

It not only fixes the "-pipe" bootstrap issue,
but also fixes the immediate ICE on invoking
cc1, etc. without an input filename (PR11581).

Thank you.


2003-10-22 Andrew Haley <>

* toplev.c (output_file_directive): Allow for null input_name.

Brings "-pipe" back to bootstrap, committed as obvious.

This didn't make it to the current snapshot...

Still, I think this is just hiding the problem.

main_input_filename continues to remain NULL,
though the fact doesn't seem to manifest itself
anywhere else if your bootstrap with "-pipe"

The real problem seems to be something else
as I note in the fourth paragraph of this message:

Why is the output filename given to "-o"
eaten up on the specs expanding machinery
in the presence of "-pipe"?


He's the fastest Indian.He's Narain Karthikeyan. Will he make it to F1?

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]