https://sourceware.org/bugzilla/show_bug.cgi?id=32260
--- Comment #10 from Jonny Weir <jonny.weir at clearpool dot io> --- (In reply to Michael Matz from comment #9) > So, I have two patches for you, see above. It would be nice to know that > they work independendly, so if you could first test the first patch in > isolation, > so that we can know that the error handling now works. > I have tested the first patch, I am able to build and link successfully, the ld memory usage was also a lot lower than before. > (What should happen with just the first patch is that the very-large input > section is still overestimated but then simply left alone and not > deduplicated, > but pasted into the output as is). > > And then add the second patch that throttles the over-estimation a bit, and > perhaps makes it so that your large input section is then deduplicated. Equally, applying the second patch, things build and link correctly. The resultant binary is also about 300Mb smaller (compared against the binary built with only the first patch applied). Many thanks for your efforts, this appears to be this issue resolved. If you would like me to test any further updates you may have, let me know. -- You are receiving this mail because: You are on the CC list for the bug.
