summary refs log tree commit diff stats
path: root/results/scraper/fex/168
diff options
context:
space:
mode:
Diffstat (limited to '')
-rw-r--r--results/scraper/fex/1685
-rw-r--r--results/scraper/fex/16802
-rw-r--r--results/scraper/fex/16818
-rw-r--r--results/scraper/fex/168712
4 files changed, 27 insertions, 0 deletions
diff --git a/results/scraper/fex/168 b/results/scraper/fex/168
new file mode 100644
index 000000000..7b1e2fdf8
--- /dev/null
+++ b/results/scraper/fex/168
@@ -0,0 +1,5 @@
+Resolve paths passed to VFS/HLE to canonical
+Find all VFS HLE routines,  such as 

+EmulatedFDManager::OpenAt()  and resolve path arguments to canonical form before using in map<path,fd>

+

+Currently only absolute paths that match perfectly will work.
\ No newline at end of file
diff --git a/results/scraper/fex/1680 b/results/scraper/fex/1680
new file mode 100644
index 000000000..59c95d83a
--- /dev/null
+++ b/results/scraper/fex/1680
@@ -0,0 +1,2 @@
+Implement VMA merging for our memtrack code
+Follow up from #1558, after new maps, or permission changes VMAs might be mergeable.
\ No newline at end of file
diff --git a/results/scraper/fex/1681 b/results/scraper/fex/1681
new file mode 100644
index 000000000..8fd62462a
--- /dev/null
+++ b/results/scraper/fex/1681
@@ -0,0 +1,8 @@
+std::shared_mutexes can't be safelly locked across forks
+I run into this as part of #1558.

+

+If they are locked before fork, then after fork they may or may not lockup on next lock, semi-randomly.

+

+Looks like we need to roll our own `shared_mutex` implementation.

+

+In general in need a gameplan for fork safety of FEXCore / Syscalls code
\ No newline at end of file
diff --git a/results/scraper/fex/1687 b/results/scraper/fex/1687
new file mode 100644
index 000000000..cba9917de
--- /dev/null
+++ b/results/scraper/fex/1687
@@ -0,0 +1,12 @@
+shared_mutex with unique lock priority
+Having a shared lock where unique locks take priority would be beneficial.

+std::shared_mutex gives shared locks priority when a shared lock is already held.

+

+With a new lock with unique priority it should have the properties:

+

+1. When a shared lock is already held and then a unique lock is attempting to be made

+  - Subsequent `try_lock_shared` should immediately return false even though unique_lock isn't held yet

+  - Subsequent `lock_shared` should block until unique lock is granted and relinquished

+  - If a subsequent unique lock is pending while a unique lock is already held then a shared lock shouldn't be able to win the contention.

+

+Might be worth using `WFE/WFET` for faster userspace checking of lock if it is expected to not be contended for too long.
\ No newline at end of file