Path: csiph.com!v102.xanadu-bbs.net!xanadu-bbs.net!feeder.erje.net!eu.feeder.erje.net!news.stack.nl!aioe.org!.POSTED!not-for-mail From: "Rod Pemberton" Newsgroups: comp.lang.misc Subject: Re: Testers wanted! Date: Mon, 05 May 2014 04:13:08 -0400 Organization: Aioe.org NNTP Server Lines: 38 Message-ID: References: NNTP-Posting-Host: CNsg4fVcCsvs3UaOgZtQCw.user.speranza.aioe.org Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1; format=flowed; delsp=yes Content-Transfer-Encoding: 7bit X-Complaints-To: abuse@aioe.org User-Agent: Opera Mail/12.16 (Linux) X-Notice: Filtered by postfilter v. 0.8.2 Xref: csiph.com comp.lang.misc:2179 On Mon, 05 May 2014 03:07:09 -0400, Nils M Holm wrote: > Rod Pemberton wrote: >> So, what did you change from April 25th to May 4th? ... > > Thanks for the suggestion, but I think this is the wrong path. > One of the testers wrote me in private and told me that he could > reproduce that linking the *same* *objects* with GNU ld twice in > a row generates differing binaries. > Ok. But, it seems to be consistend for both here. > And even if it did not, a compiler bug would not answer the > question why the assembly output for stages 2 and 3 is identical > (while the binaries are not). My point was that something changed the expected result. It's possible, that it can be found if you work from good results to bad. I.e., a second option instead of attempting to use logic to find and fix the problem, is to locate the point at which things changed. >> Are all the changed addresses in the binaries different by the >> same amount? (I haven't checked.) I.e., fixed quantity of bytes? > > Good question! I will have a look. > A bunch from 'hexdump -C' differences seem to be 120 bytes different. A 'readelf -a' of the program headers gives different load location results. The difference for readelf is 132 bytes. scc has a value of 0x000176ee for one of it's load locations, while scc1 has 0x00017766. I'm not that familiar with readelf, yet, so it may have more information. Rod Pemberton