 Ryan Fleury
|
d723a00979
|
zero precise match info before doing matches in dbgi async match path, so we don't keep stale ones around too long, once they've been invalidated
|
2025-06-27 12:52:13 -07:00 |
|
|
Ryan Fleury
|
09a047feee
|
zero match store dbgi/idx selection info if params hash has changed; we can still return the section kind (e.g. for syntax highlighting), but we no longer want to try to refer to any specific debug info slot, since it may have been evicted.
|
2025-06-27 12:36:27 -07:00 |
|
|
Ryan Fleury
|
b315bf6811
|
switch eval debug info symbol name resolution to using async match store
|
2025-06-27 11:58:04 -07:00 |
|
|
Ryan Fleury
|
a702e187ef
|
expand match store cache to include all match info
|
2025-06-27 11:24:59 -07:00 |
|
|
Ryan Fleury
|
36f5597d8e
|
eliminate unneeded path-renormalization in debug info cache lookup
|
2025-06-25 10:52:29 -07:00 |
|
|
Ryan Fleury
|
e2eb1cd6d6
|
do not unnecessarily block on parses when closing modules
|
2025-06-25 09:41:48 -07:00 |
|
|
Ryan Fleury
|
bee2570d39
|
fix double-load in radbin conversion path; fix incorrect mutation of potentially-nil trees when snapping to a file without a specific line; fix di_close path when work is occurring
|
2025-06-18 09:56:47 -07:00 |
|
|
Ryan Fleury
|
401f117653
|
adjust step trap net build in case of memory read failure - ensure we get correct memory read, otherwise we can generate bad trap nets during heavy load. allow high vs. low priority in async rdi loading; use low-priority in pre-emptive case
|
2025-06-18 08:36:41 -07:00 |
|
|
Ryan Fleury
|
0e414a3dfb
|
fix clang build; adjust raddbg --convert option to --bin
|
2025-06-10 14:39:09 -07:00 |
|
|
Ryan Fleury
|
79bf3d8639
|
eliminate standalone programs for pdb -> rdi; dwarf -> rdi; pdb -> breakpad; sketch out 'radbin', unified cli frontend for dumping/converting/etc., plug in pdb -> rdi, pdb -> breakpad, & stub out dumper/dwarf
|
2025-06-09 15:40:48 -07:00 |
|
|
Ryan Fleury
|
94b969f203
|
copyright year
|
2025-06-02 14:13:57 -07:00 |
|
|
Ryan Fleury
|
0a51de094d
|
switch to stripe-cv-based waiting mechanism on nonzero scope touches
|
2025-05-18 12:58:44 -07:00 |
|
|
Ryan Fleury
|
f99dd2a477
|
clean up search data commit on scope-hold mechanism
|
2025-05-17 13:33:43 -07:00 |
|
|
Ryan Fleury
|
f2bf8ccb2e
|
layer color coverage / usage in thread colors
|
2025-05-09 11:15:00 -07:00 |
|
|
Ryan Fleury
|
2b1e286aad
|
preserve dbgi fuzzy searching results based on key alone, rather than throwing away all results if params change (although still re-querying if params change); also fix not clearing the autocompletion string gathered for a UI build across frames
|
2025-05-06 09:31:16 -07:00 |
|
|
Ryan Fleury
|
a5d6e8a335
|
make eval ctxs more robust to missing info, e.g. in the test harness case
|
2025-03-31 13:48:19 -07:00 |
|
|
Ryan Fleury
|
3c8f58cb38
|
add tracking to dbgi scopes
|
2025-02-21 13:23:12 -08:00 |
|
|
Ryan Fleury
|
a40c26a0eb
|
fix incorrect singly-linked-list/doubly-linked-list mismatch in dbgi search cache
|
2024-11-19 10:41:22 -08:00 |
|
|
Ryan Fleury
|
cb5f9a2e06
|
robustify search item usage code to changed sets of dbgis
|
2024-11-18 11:42:43 -08:00 |
|
|
Ryan Fleury
|
846cfd259c
|
fix incorrect search joining logic, in cases where a key has been evicted
|
2024-11-18 10:24:02 -08:00 |
|
|
Ryan Fleury
|
32cfe28094
|
exclusively lock search stripe when taking task & bumping work refcount
|
2024-11-18 09:58:34 -08:00 |
|
|
Ryan Fleury
|
f1c5c97a1b
|
fix stale detection in background fuzzy search results
|
2024-11-15 14:53:03 -08:00 |
|
|
Ryan Fleury
|
6be2792659
|
search cache eviction
|
2024-11-15 14:40:05 -08:00 |
|
|
Ryan Fleury
|
48c7bb788e
|
fix incorrect completion parameters hash storage in dbgi matching work, pt. 2
|
2024-11-15 11:15:25 -08:00 |
|
|
Ryan Fleury
|
080a001183
|
fix incorrect completion parameters hash storage in dbgi matching work
|
2024-11-15 11:12:58 -08:00 |
|
|
Ryan Fleury
|
ce90b63079
|
oops........
|
2024-11-14 10:58:52 -08:00 |
|
|
Ryan Fleury
|
6cf9b82857
|
strip out completion counter from dbgi parsing layer
|
2024-11-14 10:47:45 -08:00 |
|
|
Ryan Fleury
|
cefbd40e2c
|
fix rw lock r/w mismatch
|
2024-11-14 10:26:51 -08:00 |
|
|
Ryan Fleury
|
420f089fda
|
eliminate incorrect cases of async_push_work, where pushes which could occur on worker threads may try to write to a cache that is currently being held, when the push occurs (when all threads are saturated) - async_push_work must never be called from within a cache locking mechanism
|
2024-11-14 09:55:03 -08:00 |
|
|
Ryan Fleury
|
ff45952f07
|
re-introduce timeout-based re-requests for dbgi & fs layers; further adjust pre-emptive conversion heuristic
|
2024-11-14 09:29:01 -08:00 |
|
|
Ryan Fleury
|
fa4f9621b0
|
adjust pre-emptive conversion heuristics; fix deadlock scenario in match work
|
2024-11-14 09:10:07 -08:00 |
|
|
Ryan Fleury
|
4c0ebc851c
|
async priorities, strip out incorrect ring position calculations
|
2024-11-13 18:23:18 -08:00 |
|
|
Ryan Fleury
|
dcec3c07d9
|
pass over pre-emptive debug info conversion heuristic, to be friendlier to ue-style root project directories, with what feels like 1000s and 1000s of sub-directories with DLLs hidden in various folders
|
2024-11-13 14:40:08 -08:00 |
|
|
Ryan Fleury
|
ff90a90744
|
fix lru eviction in async name match store
|
2024-11-12 14:36:40 -08:00 |
|
|
Ryan Fleury
|
e95afcd5aa
|
fix ring buffer stomping in async match
|
2024-11-12 13:20:42 -08:00 |
|
|
Ryan Fleury
|
9808157f45
|
switch match tasks to atomically update match nodes, rather than push results into a queue
|
2024-11-12 13:00:23 -08:00 |
|
|
Ryan Fleury
|
0ec6b8e8a8
|
more work on match store experiment; consume ring buffer results on begin - this experiment is not exactly the right approach, though. because the ui thread definitionally learns about these identifiers faster than the matches can (always) be completed, there is a permanent writer-going-faster-than-reader problem. the correct approach will need to account for this.
|
2024-11-12 13:00:23 -08:00 |
|
|
Ryan Fleury
|
3e0c150ccb
|
more work on background name matching & plugging it into syntax highlighting path
|
2024-11-12 13:00:23 -08:00 |
|
|
Ryan Fleury
|
420c2ec208
|
fix dbgi key array copy
|
2024-11-11 16:15:15 -08:00 |
|
|
Ryan Fleury
|
f9abd3efb9
|
first half of first pass at asynchronous name matching; simplify current name matching path used in syntax highlighting & deduplicate between code slice & code labels
|
2024-11-11 16:06:01 -08:00 |
|
|
Ryan Fleury
|
5893ab94cb
|
cancellation mechanism for debug info fuzzy searching
|
2024-11-11 10:38:52 -08:00 |
|
|
Ryan Fleury
|
ff2e9df705
|
fix incorrect match range list copying in new searcher threads
|
2024-11-11 10:02:06 -08:00 |
|
|
Ryan Fleury
|
a89af7f6fd
|
scratch that - *now* fix the arrayification
|
2024-11-07 10:07:36 -08:00 |
|
|
Ryan Fleury
|
5a19ab9c31
|
fix new search path arrayification
|
2024-11-07 10:06:49 -08:00 |
|
|
Ryan Fleury
|
e204fe28c1
|
eliminate deadlocking mechanism in dbgi search cache lookup/mutation; replace with simpler but possibly worse mechanism for now
|
2024-11-07 09:57:25 -08:00 |
|
|
Ryan Fleury
|
d036eb9fc5
|
eliminate old dbgi_search layer
|
2024-11-05 15:08:58 -08:00 |
|
|
Ryan Fleury
|
1621e504ce
|
first half pass at new dbgi fuzzy searching functionality; distributes search amongst sub-tasks
|
2024-11-05 15:07:06 -08:00 |
|
|
Ryan Fleury
|
4a0e21d447
|
collapse user clock indices into base-layer-provided clock counter; fuzzy_search -> dbgi_search
|
2024-11-04 13:12:52 -08:00 |
|
|
Ryan Fleury
|
c3a672cd02
|
fix incorrect async work ordering in dbgi layer
|
2024-11-03 12:48:34 -08:00 |
|
|
Ryan Fleury
|
a40edbba51
|
dasm, ctrl memstream, & more moving to async layer, rather than per-layer thread pools
|
2024-11-03 12:11:24 -08:00 |
|