New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
should cmovcc and fmovcc have dst as a src? #269
Comments
From bruen...@google.com on June 06, 2013 18:02:26 xref issue #1181 Owner: --- |
From bruen...@google.com on November 06, 2013 09:32:14 OP_getsec also has conditional writes to destinations due to its dynamic leaf functions. For now I'm listing these destinations as both sources and dests. Xref issue #1313 . |
From bruen...@google.com on November 06, 2013 10:36:14 OP_xend also has a conditional write to eax, again being modeled as a source and dest. |
From bruen...@google.com on November 08, 2013 15:09:39 OP_vpmaskmovd and OP_vpmaskmovd have memory store forms that conditionally write to memory, which is again modeled with the memory as both a source and a dest. |
From bruen...@google.com on October 07, 2014 20:17:30 ** TODO for arm 32-bit we have a lot of predication => first-class support We're thinking that we'll add predication info to instr_t.prefixes, and add Should we remove the x86 dst-is-src trick we added to fool liveness analysis? Owner: bruen...@google.com |
From bruen...@google.com on October 24, 2014 07:54:37 We plan to remove the dst-is-src. But I'm still not sure what the best predication approach is for |
From bruen...@google.com on October 24, 2014 08:29:51 For composability of operations, and matching iteration, keeping it as "might And file an issue on removing all predication during mangling, under an option? |
From derek.br...@gmail.com on October 27, 2014 13:17:51 This issue was closed by revision r2914 . Status: Fixed |
From derek.br...@gmail.com on March 08, 2010 20:53:19
cmovcc in DR's IR lists the dst as a src, in order to avoid code analysis
from treating the dst as a def since it's not always written: not a perfect
solution but it can avoid special-casing by typical analysis code. should
I leave that in there or is it misleading? isn't it misleading to not have
it in there? should fcmovcc do the same (currently it does not have dst as
src)?
Original issue: http://code.google.com/p/dynamorio/issues/detail?id=269
The text was updated successfully, but these errors were encountered: