diff options
| author | Christian Krinitsin <mail@krinitsin.com> | 2025-07-17 09:10:43 +0200 |
|---|---|---|
| committer | Christian Krinitsin <mail@krinitsin.com> | 2025-07-17 09:10:43 +0200 |
| commit | f2ec263023649e596c5076df32c2d328bc9393d2 (patch) | |
| tree | 5dd86caab46e552bd2e62bf9c4fb1a7504a44db4 /results/scraper/fex/3948 | |
| parent | 63d2e9d409831aa8582787234cae4741847504b7 (diff) | |
| download | emulator-bug-study-f2ec263023649e596c5076df32c2d328bc9393d2.tar.gz emulator-bug-study-f2ec263023649e596c5076df32c2d328bc9393d2.zip | |
add downloaded fex bug-reports
Diffstat (limited to 'results/scraper/fex/3948')
| -rw-r--r-- | results/scraper/fex/3948 | 11 |
1 files changed, 11 insertions, 0 deletions
diff --git a/results/scraper/fex/3948 b/results/scraper/fex/3948 new file mode 100644 index 00000000..979fa0ab --- /dev/null +++ b/results/scraper/fex/3948 @@ -0,0 +1,11 @@ +Cross-process tear free split-lock emulation can't be implemented without kernel support (or TME) +In order to correctly handle split-lock emulation, we need a global mutex. We could easily get this through FEXServer handing a shared page to all FEXInterpreter instances. + +The problem with this is that if a process dies when one of its threads is handling a split-lock (Or cross-16byte boundary atomic) then the mutex will be permanently set. Effectively hanging every running FEXInterpreter process on the system. + +There's no way to work around this in a tear-free fashion and also never have hangs. It /needs/ to be done in the kernel. + +The "alternative" and preferred approach would be if all ARM vendors just support TME, but that's never going to occur. + +As for starting the discussion with the ARM kernel maintainers for getting this supported upstream... + |