Commit Graph

70 Commits

Author SHA1 Message Date
David Harvey
7b3f703047 platform extensions can enable tracelog by default (it can still be overriden in Target.GlobalDefinitions). probably should become a project setting at some point!
#jira UE-88925
#rb jon.cain
#fyi mickael.gilabert, martin.ridgers

[CL 11536119 by David Harvey in 4.25 branch]
2020-02-19 08:06:48 -05:00
Rolando Caloca
5b82f15def Copying //UE4/Dev-RenderPlat-Staging@11388153 to //UE4/Main
#rb none
#rnx

[CL 11388545 by Rolando Caloca in Main branch]
2020-02-12 13:27:19 -05:00
Martin Ridgers
44580d01d4 Fixed HoloLens arm64 compile error.
#rb trivial
#rnx
#jira UE-88508


#ushell-cherrypick of 11373497 by Martin.Ridgers

[CL 11373516 by Martin Ridgers in Main branch]
2020-02-12 02:40:15 -05:00
Stefan Boberg
55a8b15f1f Copying //UE4/Dev-Core to Main (//UE4/Main)
#rb none

[CL 11258477 by Stefan Boberg in Main branch]
2020-02-05 14:26:36 -05:00
Martin Ridgers
fa37399e19 Do not allow sockets to be inherited. Child processes spawned with UE's CreateProc() are set to inherit their parent's handles. This results in listening sockets lingering after the ower's terminated if its children still a reference to these sockets (i.e. they were inherited). As these lingering sockets can still be connected to, no end of odd socket behaviour ensues.
#rb none
#rnx

[CL 11152126 by Martin Ridgers in Dev-Core branch]
2020-01-29 02:21:24 -05:00
Martin Ridgers
8b43ac0bb0 The initialisation of thread-local variables is unsequenced with regard to other translation units which meant TlsContext might not be initialised in time. So now the thread-local buffer pointer and the "null" buffer are constant-initialised variables which will definitely have suitable values before any code is run. TlsContext is still required to detect when threads exit so we don't leak buffers so it's used to get a thread ID to guarantee TlsContext exists and is initialised (compilers are free to lazy-initialise if they so wish so it must be used at least once). Long message this isn't it? No one will read this far.
#rb none
#rnx

[CL 11115075 by Martin Ridgers in Dev-Core branch]
2020-01-27 09:07:25 -05:00
CarlMagnus Nordin
e66592a12f Compile fixes
#rb none

[CL 11114810 by CarlMagnus Nordin in Dev-Core branch]
2020-01-27 07:22:41 -05:00
Martin Ridgers
8c64e6ab81 Moved the runtime-side command/control handling out into its own translation unit.
#rb trivial
#rnx

[CL 11114543 by Martin Ridgers in Dev-Core branch]
2020-01-27 02:57:55 -05:00
PJ Kack
ef8a7b620e AUTO MERGE
Merging //UE4/Main @ 11112898 to //UE4/Dev-Core (dev-core-l0178)

#rb none

(ushell-p4-mergedown)

[CL 11113593 by PJ Kack in Dev-Core branch]
2020-01-26 11:35:42 -05:00
PJ Kack
3a9ca4d2f9 Merging //UE4/Main @ 11112898 to //UE4/Dev-Core (dev-core-l0178)
#rb none

(ushell-p4-mergedown)

[CL 11113592 by PJ Kack in Dev-Core branch]
2020-01-26 11:33:22 -05:00
Martin Ridgers
e643f524ac Organise trace event buffers into lists per thread. This allows us to only send the oldest of a thread's event data first, and never process future buffers until prior ones are complete.
#rb none
#rnx

[CL 11113562 by Martin Ridgers in Dev-Core branch]
2020-01-26 09:19:40 -05:00
Martin Ridgers
5bc4fba5a0 Do not keep a thread ID in the TlsContext object so its machine-word size.
#rb none
#rnx

[CL 11113555 by Martin Ridgers in Dev-Core branch]
2020-01-26 09:04:46 -05:00
Martin Ridgers
b0adc120ec Use a zero-valued thread ID as an indication that there's not valid trace event buffer yet.
#rb none
#rnx

[CL 11113544 by Martin Ridgers in Dev-Core branch]
2020-01-26 08:50:16 -05:00
Martin Ridgers
85a04eec01 Extracted the Send() lambda into its own function to aid ConsomeEvents' complexity.
#rb none
#rnx

[CL 11113525 by Martin Ridgers in Dev-Core branch]
2020-01-26 08:31:18 -05:00
Martin Ridgers
a9f3b0023a Use increment instead of a CAS loop.
#rb none
#rnx

[CL 11113516 by Martin Ridgers in Dev-Core branch]
2020-01-26 08:07:52 -05:00
Martin Ridgers
c43f5a655e If an the header for a blob of auxilary (array) data ended up in the overflow a commit could get missed causing buffers of trace data to stop draining.
#rb none
#rnx

[CL 11113511 by Martin Ridgers in Dev-Core branch]
2020-01-26 07:43:28 -05:00
Martin Ridgers
10f6b798c5 Always send important events for now otherwise we miss channel announcements.
#rb jb
#rnx

[CL 11095800 by Martin Ridgers in Dev-Core branch]
2020-01-23 12:09:26 -05:00
Martin Ridgers
2a318368dc Compile-time access to a trace event's flags.
#rb jb
#rnx

[CL 11091786 by Martin Ridgers in Dev-Core branch]
2020-01-23 07:25:57 -05:00
Martin Ridgers
7c2cf042a8 Only initialise a trace event if we are going to be sending it.
#rb none
#rnx

[CL 11080205 by Martin Ridgers in Dev-Core branch]
2020-01-22 03:08:31 -05:00
Martin Ridgers
ddfd315661 Only channels should dictate if an event is sent or not.
#rb none
#rnx

[CL 11080200 by Martin Ridgers in Dev-Core branch]
2020-01-22 03:06:02 -05:00
Martin Ridgers
6c6fcfdf0c Removed event hashes and the linked list.
#rb none
#rnx

[CL 11080032 by Martin Ridgers in Dev-Core branch]
2020-01-22 02:34:03 -05:00
Johan Berg
6596dc076d Removse event enable/disable.
#rb martin.ridgers

[CL 11072617 by Johan Berg in Dev-Core branch]
2020-01-21 10:50:50 -05:00
Johan Berg
fa91465163 Remove use of "Always" from inactive code blocks.
#rb trivial

[CL 11066680 by Johan Berg in Dev-Core branch]
2020-01-20 14:36:54 -05:00
Martin Ridgers
1e4e73396e Added missing null terminator.
#rb trivial
#rnx

[CL 11062153 by Martin Ridgers in Dev-Core branch]
2020-01-19 08:04:00 -05:00
michael noland
2d6847d01a Core: Fixes for various unsafe implicit type conversion warnings in the Core module (WIP; some files still suppress the warning)
- Ptrdiff -> int32
- Float/int confusion and double/float
- size_t stuff; various changes to the algorithms to use a deduced IndexType template argument and/or decltype to use the appropriate size for indicies and counts
- Fixed GetNum(FString) incorrectly returning SIZE_T instead of int32, and GetNum(container) now returns whatever container.Num() does (so usually int32)
#jira UE-86949
#rb marc.audy, steve.robb


#ROBOMERGE-OWNER: michael.noland
#ROBOMERGE-AUTHOR: michael.noland
#ROBOMERGE-SOURCE: CL 11050799 via CL 11050828 via CL 11050837
#ROBOMERGE-BOT: (v637-11041722)

[CL 11051763 by michael noland in Main branch]
2020-01-17 13:49:41 -05:00