mirror of
https://github.com/mihawk-99/PS5_RetroArch.git
synced 2026-10-18 23:09:29 -05:00
Patch 0093: release the HDR uniform buffer on every video free
vulkan_init creates vk->hdr.ubo whenever the build has VULKAN_HDR_SWAPCHAIN, but vulkan_free destroyed it only when the context reported HDR support, which this console's surface never does. One 256-byte host-visible buffer leaked each video init, two per core reload: the two direct-memory mappings a reload that Profile 9 left unidentified. The patch count goes from 202 to 203 for this one block. Title 9a84b116, four reloads of Melee: every reload returns to 180 live mappings and memory:152 (evidence/dolphin-p9-reload-flat). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
MihawkandClaude Opus 5.5 committed
1 parent
35c905b6dd
commit
18689a115f
6 files changed
+103
-1
No files matched your search
@@ -3501,3 +3501,22 @@ CPU's values win, as a CPU write does on the hardware; a gather-pipe burst, whic
|
||||
only adds, is retried. Title 08e8ed4c, twice from boot for seven minutes (368
|
||||
emulated seconds, past the old failure and through a second attract loop): no
|
||||
alert, no unknown opcode, no crash.
|
||||
|
||||
## 2026-09-25 — Profile 9's last two mappings: the HDR uniform buffer
|
||||
|
||||
Profile 9 left two direct-memory mappings a reload unidentified. A bounded
|
||||
one-shot registry in ../PS5_Vulkan's allocator (removed since; title 1ecbfd3a)
|
||||
recorded the caller of every live VkDeviceMemory. After each reload, one entry grew
|
||||
by 2 (1, 3, 5, 7, 9 over four reloads): 256 bytes of host-visible memory from
|
||||
`vulkan_create_buffer`, called by `vulkan_init` for `vk->hdr.ubo`. RetroArch
|
||||
creates that uniform buffer whenever the build has `VULKAN_HDR_SWAPCHAIN`, but
|
||||
`vulkan_free` destroyed it only when the context has HDR support, which this
|
||||
console's surface never reports. The video driver is initialised twice a reload,
|
||||
so two buffers leaked each time.
|
||||
|
||||
Patch 0093 releases the buffer whenever it exists; the HDR images stay with the
|
||||
support flag that creates them. Title 9a84b116, the same four-reload Melee run
|
||||
(`evidence/dolphin-p9-reload-flat`): every reload returns to `direct_live=180`
|
||||
and `memory:152`, the count the first boot held. Before 0093, memory grew 152,
|
||||
154, 156, 158, 160. The stage and table mappings return to the same counts each
|
||||
cycle. There were no crashes, and each steady window ran at 97-100%.
|
||||
@@ -0,0 +1,38 @@
|
||||
{
|
||||
"step": "dolphin-p9-reload-flat",
|
||||
"recorded": "2026-09-25T19:25:07Z",
|
||||
"console_run": "",
|
||||
"revision": "title 9a84b116: driver 2f1e447, RetroArch patch 0093 on 11f5a81",
|
||||
"command": "tooling/dolphin-profiles/p1-accurate.txt; Melee from my entry state; the pad script's four Quick Menu closes and RELOADs; driver profile armed",
|
||||
"raw_capture": "klog/p9-0093-reload-excerpt.txt",
|
||||
"raw_lines": 26,
|
||||
"raw_bytes": 2900,
|
||||
"lines": [
|
||||
"build identity: 9a84b116b5320eda8bdd2e609a6f1db51d47335d29d9212b5ea0ef01d86a1c61",
|
||||
"[ps5vk] profile ... direct_live=266/601MiB ... direct_kinds=memory:152,stage:62,compute:11,tables:37,query:1,queue:2,display:1",
|
||||
"[ps5vk] profile ... direct_live=271/601MiB ... direct_kinds=memory:152,stage:67,compute:11,tables:37,query:1,queue:2,display:1",
|
||||
"[ps5vk] profile ... direct_live=272/601MiB ... direct_kinds=memory:152,stage:68,compute:11,tables:37,query:1,queue:2,display:1",
|
||||
"[ps5vk] profile ... direct_live=376/610MiB ... direct_kinds=memory:256,stage:70,compute:11,tables:35,query:1,queue:2,display:1",
|
||||
"[ps5vk] profile ... direct_live=298/604MiB ... direct_kinds=memory:152,stage:87,compute:11,tables:44,query:1,queue:2,display:1",
|
||||
"[ps5vk] profile ... direct_live=305/605MiB ... direct_kinds=memory:152,stage:94,compute:11,tables:44,query:1,queue:2,display:1",
|
||||
"[ps5vk] profile ... direct_live=418/613MiB ... direct_kinds=memory:256,stage:105,compute:11,tables:42,query:1,queue:2,display:1",
|
||||
"input: pad script RELOAD at 82.66 s: 1",
|
||||
"[ps5vk] profile ... direct_live=180/592MiB ... direct_kinds=memory:152,stage:10,compute:1,tables:13,query:1,queue:2,display:1",
|
||||
"[ps5vk] profile ... direct_live=183/592MiB ... direct_kinds=memory:152,stage:12,compute:2,tables:13,query:1,queue:2,display:1",
|
||||
"[ps5vk] profile ... direct_live=272/601MiB ... direct_kinds=memory:152,stage:69,compute:10,tables:37,query:1,queue:2,display:1",
|
||||
"input: pad script RELOAD at 133.81 s: 1",
|
||||
"[ps5vk] profile ... direct_live=180/592MiB ... direct_kinds=memory:152,stage:10,compute:1,tables:13,query:1,queue:2,display:1",
|
||||
"[ps5vk] profile ... direct_live=183/592MiB ... direct_kinds=memory:152,stage:12,compute:2,tables:13,query:1,queue:2,display:1",
|
||||
"[ps5vk] profile ... direct_live=273/601MiB ... direct_kinds=memory:152,stage:70,compute:10,tables:37,query:1,queue:2,display:1",
|
||||
"input: pad script RELOAD at 184.28 s: 1",
|
||||
"[ps5vk] profile ... direct_live=180/592MiB ... direct_kinds=memory:152,stage:10,compute:1,tables:13,query:1,queue:2,display:1",
|
||||
"[ps5vk] profile ... direct_live=183/592MiB ... direct_kinds=memory:152,stage:12,compute:2,tables:13,query:1,queue:2,display:1",
|
||||
"[ps5vk] profile ... direct_live=273/601MiB ... direct_kinds=memory:152,stage:70,compute:10,tables:37,query:1,queue:2,display:1",
|
||||
"input: pad script RELOAD at 234.68 s: 1",
|
||||
"[ps5vk] profile ... direct_live=180/592MiB ... direct_kinds=memory:152,stage:10,compute:1,tables:13,query:1,queue:2,display:1",
|
||||
"[ps5vk] profile ... direct_live=183/592MiB ... direct_kinds=memory:152,stage:12,compute:2,tables:13,query:1,queue:2,display:1",
|
||||
"[ps5vk] profile ... direct_live=273/601MiB ... direct_kinds=memory:152,stage:70,compute:10,tables:37,query:1,queue:2,display:1",
|
||||
"[ps5vk] profile ... direct_live=281/602MiB ... direct_kinds=memory:152,stage:78,compute:10,tables:37,query:1,queue:2,display:1",
|
||||
"[ps5vk] profile ... direct_live=295/603MiB ... direct_kinds=memory:152,stage:92,compute:10,tables:37,query:1,queue:2,display:1"
|
||||
]
|
||||
}
|
||||
@@ -0,0 +1,14 @@
|
||||
{
|
||||
"step": "dolphin-p9-reload-flat",
|
||||
"note": "After 0093: every reload returns to direct_live=180 and memory:152, the same count the first boot held; the stage and table mappings return to 10/12/70 each cycle. Before it, memory grew 152, 154, 156, 158, 160 (two per reload, the HDR uniform buffer). Four reloads, no crash.",
|
||||
"must_contain": [
|
||||
"input: pad script RELOAD at 234.68 s: 1",
|
||||
"direct_kinds=memory:152,stage:10"
|
||||
],
|
||||
"must_not_contain": [
|
||||
"crash:",
|
||||
"direct_kinds=memory:154",
|
||||
"direct_kinds=memory:160"
|
||||
],
|
||||
"human_check": ""
|
||||
}
|
||||
@@ -168,3 +168,8 @@
|
||||
# were overwritten by the wide ones and leaked on every re-initialisation
|
||||
# gfx/drivers/vulkan.c
|
||||
0092-release-replaced-textures
|
||||
|
||||
# 0093 release the HDR uniform buffer whenever vulkan_init created it, not only
|
||||
# when the context reports HDR support: it leaked on every video init
|
||||
# gfx/drivers/vulkan.c
|
||||
0093-release-hdr-uniform-buffer
|
||||
@@ -342,7 +342,7 @@ class PortPatches(unittest.TestCase):
|
||||
change deliberately, in a commit that says why.
|
||||
"""
|
||||
self.assertEqual(
|
||||
len(self.blocks()), 202,
|
||||
len(self.blocks()), 203,
|
||||
"the patch count changed: if a block was added or removed on purpose, "
|
||||
"update this number in the same commit and say why in its message")
|
||||
|
||||
|
||||
@@ -2914,6 +2914,32 @@ EDITS = [
|
||||
" vulkan_destroy_texture(vk->context->device, &vk->display.blank_texture);\n",
|
||||
"patches/series, 0092): the 4x4 texture",
|
||||
),
|
||||
(
|
||||
# 0093: vulkan_init creates the HDR uniform buffer whenever the build
|
||||
# has VULKAN_HDR_SWAPCHAIN, but vulkan_free released it only when the
|
||||
# context had HDR support, which this console's surface never reports.
|
||||
# One 256-byte host-visible buffer and its memory leaked on every video
|
||||
# driver init: two a core reload, seen as caller vulkan_create_buffer
|
||||
# in the driver's allocation registry after 0092 (2026-09-25). The
|
||||
# buffer is released whenever it exists; the HDR images stay with the
|
||||
# support flag that created them.
|
||||
"gfx/drivers/vulkan.c",
|
||||
"#ifdef VULKAN_HDR_SWAPCHAIN\n"
|
||||
" if (vk->context->flags & VK_CTX_FLAG_HDR_SUPPORT)\n"
|
||||
" {\n"
|
||||
" vulkan_destroy_buffer(vk->context->device, &vk->hdr.ubo);\n"
|
||||
" vulkan_destroy_hdr_buffer(vk->context->device, &vk->main_buffer);\n",
|
||||
"#ifdef VULKAN_HDR_SWAPCHAIN\n"
|
||||
" /* Changed by this port (patches/series, 0093): vulkan_init creates the\n"
|
||||
" * uniform buffer with or without HDR support, so it is released the\n"
|
||||
" * same way. */\n"
|
||||
" if (vk->hdr.ubo.memory != VK_NULL_HANDLE)\n"
|
||||
" vulkan_destroy_buffer(vk->context->device, &vk->hdr.ubo);\n"
|
||||
" if (vk->context->flags & VK_CTX_FLAG_HDR_SUPPORT)\n"
|
||||
" {\n"
|
||||
" vulkan_destroy_hdr_buffer(vk->context->device, &vk->main_buffer);\n",
|
||||
"patches/series, 0093): vulkan_init creates the",
|
||||
),
|
||||
]
|
||||
|
||||
# Changes that are withdrawn rather than deleted, by marker.
|
||||
|
||||
Reference in new issue
Block a user