On 9/14/06, <b class="gmail_sendername">Olaf Hering</b> <<a href="mailto:olh@suse.de">olh@suse.de</a>> wrote:<div><span class="gmail_quote"></span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
The current trunk gives these make check failures in openSuSE 10.2 beta.<br>ltrace -x does not work, on ppc64 at least.</blockquote><div><br>ia64 is broken too. <br></div><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
FAIL: func_intptr_ret(42) in ./ltrace.main/parameters.ltrace for 0 times ,should be 1</blockquote><div><br>I see you have this failure on powerpc-suse-linux-gnu. On ia64, the fix for that was in gimme_arg when reading parameters after the function had been returned. It was assuming that the register renaming state was the same during and after a function call. Does PPC do something similar?
<br><br>An alternative fix would be to remember the parameter values while still inside the function call.<br></div><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
FAIL: func_strlen(\"Hello world\") *= *11 in ./ltrace.main/parameters.ltrace for 0 times ,should be 1</blockquote><div><br>Same problem. <br><br></div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
FAIL: func_float(3.40*, -3.40*) in ./ltrace.main/parameters.ltrace for 0 times ,should be 1</blockquote><div><br>On ia64, this was due to floating point parameters being passed in FP registers. The patch I just posted will not fix the PPC problem, but something similar might be needed.
<br></div><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">FAIL: func_struct({ 89, 2, 3, . { 1, 10 }, { 3, 30 } ., . { 4, 40 }, { 5, 50 }, { 6, 60 } ., \"123\" }) in ./ltrace.main/parameters.ltrace for 0 times ,should be 1
</blockquote><div><br>I was not handling alignment properly in several cases. My recent patch may fix this one.<br><br></div><blockquote>FAIL: func_intptr(17) in ./ltrace.main/parameters.ltrace for 0 times ,should be 1</blockquote>
<div>I haven't seen this one. <br></div><br><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">FAIL: func_arrayf(. 10.10*, 11.10*, 12.10*,
13.10*\.\.\. ., ) in ./ltrace.main/parameters.ltrace for 0 times ,should be 1<br></blockquote><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
FAIL: func_arrayf(. 10.10*, 11.10* ., ) in ./ltrace.main/parameters.ltrace for 0 times ,should be 1</blockquote><div><br>I had these on ia64 after I fixed other things. It was a double vs float problem. My patch tweaks how this is done, but probably will not fix this one.
<br></div><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Running ./ltrace.minor/count-record.exp ...<br>FAIL: Fail to find 2 remove in ./ltrace.minor/count-
record.ltrace<br>FAIL: Fail to find 1 getcwd in ./ltrace.minor/count-record.ltrace<br>FAIL: Fail to find 1 fread in ./ltrace.minor/count-record.ltrace</blockquote><div><br>What does ./ltrace.minor/count-record.ltrace have?
<br></div><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Native configuration is powerpc64-suse-linux-gnu<br><br>FAIL: func_ignore(1, *, *3) in ./ltrace.main/parameters.ltrace for 0 times ,should be 1
</blockquote><div><br>Wow, this one is just unhappy. Can you post ./ltrace.main/parameters.ltrace?</div><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Native configuration is x86_64-suse-linux-gnu<br>FAIL: func_float(3.40*, -3.40*) in ./ltrace.main/parameters.ltrace for 0 times ,should be 1</blockquote><div><br>It would be nice if the testing harness could isolate sections and print out "got: XXX, expected: YYY".
<br></div><br>When I next get a chance, I'll check whether HP's testdrive cluster has any of these architectures and see what I can do about tracking the problems down. It's not straightforward, because they typically don't install elfutils-libelf-devel, which is required by ltrace.
<br><br></div>