MOTHBALLED-graphviz
MOTHBALLED-graphviz copied to clipboard
Non-determinancy in fdp
The output from running fdp on the regression test file rtests/graph/fdp.gv can change between runs. This appears unrelated to the use of drand, as the seed is always set the same in every run. There appear to be just two possible outcomes, related to order of two edges in a derived subgraph. So the question is, why does the ordering change?
OSX does Address Space Layout Randomization. https://en.wikipedia.org/wiki/Address_space_layout_randomization https://en.wikipedia.org/wiki/Address_space_layout_randomization
Programs that rely hashing or comparing addresses could be affected.
This can be disabled with the linker option no_pie http://stackoverflow.com/questions/23897963/documented-way-to-disable-aslr-on-os-x http://stackoverflow.com/questions/23897963/documented-way-to-disable-aslr-on-os-x
So for graphviz, sometimes I ./configure CFLAGS='-Wl,-no_pie -g’ when debugging.
On Mar 2, 2017, at 2:12 PM, emden [email protected] wrote:
The output from running fdp on the regression test file rtests/graph/fdp.gv can change between runs. This appears unrelated to the use of drand, as the seed is always set the same in every run. There appear to be just two possible outcomes, related to order of two edges in a derived subgraph. So the question is, why does the ordering change?
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/ellson/graphviz/issues/1214, or mute the thread https://github.com/notifications/unsubscribe-auth/ACtWzyYXL6cb7aYXTp7wBlFRWX4Ns8Ezks5rhxSagaJpZM4MRZyw.
Does this also occur on Linux, because that is where I encountered it?
Why, yes, it does, since 2005.
https://linux-audit.com/linux-aslr-and-kernelrandomize_va_space-setting/ https://linux-audit.com/linux-aslr-and-kernelrandomize_va_space-setting/
It’s been a while but I’m pretty sure I’ve done sudo echo 0 > /proc/sys/kernel/randomize_va_space
remember to set it back to 2 when you’re done debugging!
On Mar 2, 2017, at 5:09 PM, emden [email protected] wrote:
Does this also occur on Linux, because that is where I encountered it?
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/ellson/graphviz/issues/1214#issuecomment-283798562, or mute the thread https://github.com/notifications/unsubscribe-auth/ACtWz9y3uamoHrKbLp4IgtBRF7B40IpFks5rhz4pgaJpZM4MRZyw.