[PATCH] D52610: [Esan] Port cache frag to FreeBSD
krytarowski resigned from this revision. krytarowski added a comment. Herald added a project: clang. ESan is now discontinued. Repository: rC Clang CHANGES SINCE LAST ACTION https://reviews.llvm.org/D52610/new/ https://reviews.llvm.org/D52610 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D52610: [Esan] Port cache frag to FreeBSD
krytarowski added a comment. Can you try to port this mutant thread-process in ESan to FreeBSD? I don't want to see changes in generic LLVM code. Repository: rC Clang CHANGES SINCE LAST ACTION https://reviews.llvm.org/D52610/new/ https://reviews.llvm.org/D52610 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D52610: [Esan] Port cache frag to FreeBSD
devnexen added a comment. Things might differ between NetBSD and FreeBSD about the feasibility. So maybe for the former it is more reachable (only would need to set specific application mappings maybe ?). - So once the non writable addresses are created, it processes the working sets to make it world aligned => One of the points of failure. - The segfault handler does not work or can't be instrumented because not initialised yet at this stage. The caching frag works because there is no shadow mapping (yet?). Repository: rC Clang https://reviews.llvm.org/D52610 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D52610: [Esan] Port cache frag to FreeBSD
krytarowski added a comment. In https://reviews.llvm.org/D52610#1279176, @devnexen wrote: > - FreeBSD does not have real Linux's clone equivalent. > - Hangs or crashes during the final report (the shadow mapping is similar as > Linux's though). NetBSD has a clone compatibility syscall... but please try to explain what is the algorithm to map to a BSD system. There must but a way to port it to BSD. Repository: rC Clang https://reviews.llvm.org/D52610 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D52610: [Esan] Port cache frag to FreeBSD
devnexen added a comment. - FreeBSD does not have real Linux's clone equivalent. - Hangs or crashes during the final report (the shadow mapping is similar as Linux's though). Repository: rC Clang https://reviews.llvm.org/D52610 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D52610: [Esan] Port cache frag to FreeBSD
krytarowski added a comment. In https://reviews.llvm.org/D52610#1278743, @devnexen wrote: > ping working-set on FreeBSD does not seem doable. Rationale? I don't feel convinced. Repository: rC Clang https://reviews.llvm.org/D52610 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D52610: [Esan] Port cache frag to FreeBSD
devnexen added a comment. ping working-set on FreeBSD does not seem doable. Repository: rC Clang https://reviews.llvm.org/D52610 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D52610: [Esan] Port cache frag to FreeBSD
devnexen added inline comments. Comment at: lib/CodeGen/BackendUtil.cpp:323 Opts.ToolType = EfficiencySanitizerOptions::ESAN_CacheFrag; - else if (LangOpts.Sanitize.has(SanitizerKind::EfficiencyWorkingSet)) + else if (T.getOS() == Triple::Linux && +LangOpts.Sanitize.has(SanitizerKind::EfficiencyWorkingSet)) krytarowski wrote: > devnexen wrote: > > krytarowski wrote: > > > Is it possible to port it to FreeBSD and skip some conditions in generic > > > code? > > Not for now, working-set hangs at init time on FreeBSD and unit tests had > > been disabled in the compiler-rt part. > So can the support be extended in compiler-rt? I would find it more useful to > handle featured sanitizer in the first place. Not sure of the feasibility yet, let s say for now it s just a starting point. Repository: rC Clang https://reviews.llvm.org/D52610 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D52610: [Esan] Port cache frag to FreeBSD
krytarowski added inline comments. Comment at: lib/CodeGen/BackendUtil.cpp:323 Opts.ToolType = EfficiencySanitizerOptions::ESAN_CacheFrag; - else if (LangOpts.Sanitize.has(SanitizerKind::EfficiencyWorkingSet)) + else if (T.getOS() == Triple::Linux && +LangOpts.Sanitize.has(SanitizerKind::EfficiencyWorkingSet)) devnexen wrote: > krytarowski wrote: > > Is it possible to port it to FreeBSD and skip some conditions in generic > > code? > Not for now, working-set hangs at init time on FreeBSD and unit tests had > been disabled in the compiler-rt part. So can the support be extended in compiler-rt? I would find it more useful to handle featured sanitizer in the first place. Repository: rC Clang https://reviews.llvm.org/D52610 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D52610: [Esan] Port cache frag to FreeBSD
devnexen added inline comments. Comment at: lib/CodeGen/BackendUtil.cpp:323 Opts.ToolType = EfficiencySanitizerOptions::ESAN_CacheFrag; - else if (LangOpts.Sanitize.has(SanitizerKind::EfficiencyWorkingSet)) + else if (T.getOS() == Triple::Linux && +LangOpts.Sanitize.has(SanitizerKind::EfficiencyWorkingSet)) krytarowski wrote: > Is it possible to port it to FreeBSD and skip some conditions in generic code? Not for now, working-set hangs at init time on FreeBSD and unit tests had been disabled in the compiler-rt part. Repository: rC Clang https://reviews.llvm.org/D52610 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D52610: [Esan] Port cache frag to FreeBSD
krytarowski added inline comments. Comment at: lib/CodeGen/BackendUtil.cpp:323 Opts.ToolType = EfficiencySanitizerOptions::ESAN_CacheFrag; - else if (LangOpts.Sanitize.has(SanitizerKind::EfficiencyWorkingSet)) + else if (T.getOS() == Triple::Linux && +LangOpts.Sanitize.has(SanitizerKind::EfficiencyWorkingSet)) Is it possible to port it to FreeBSD and skip some conditions in generic code? Repository: rC Clang https://reviews.llvm.org/D52610 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D52610: [Esan] Port cache frag to FreeBSD
devnexen added a comment. ping. Repository: rC Clang https://reviews.llvm.org/D52610 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D52610: [Esan] Port cache frag to FreeBSD
devnexen added a comment. ping compiler-rt component committed :-) Repository: rC Clang https://reviews.llvm.org/D52610 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D52610: [Esan] Port cache frag to FreeBSD
devnexen added a comment. The compiler-rt part had been committed. Repository: rC Clang https://reviews.llvm.org/D52610 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits