forked from djasper/clang
-
Notifications
You must be signed in to change notification settings - Fork 0
/
NOTES.txt
111 lines (80 loc) · 4.28 KB
/
NOTES.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
//===---------------------------------------------------------------------===//
// Random Notes
//===---------------------------------------------------------------------===//
//===---------------------------------------------------------------------===//
To time GCC preprocessing speed without output, use:
"time gcc -MM file"
This is similar to -Eonly.
//===---------------------------------------------------------------------===//
Creating and using a PTH file for performance measurement (use a release build).
$ clang -ccc-pch-is-pth -x objective-c-header INPUTS/Cocoa_h.m -o /tmp/tokencache
$ clang -cc1 -token-cache /tmp/tokencache INPUTS/Cocoa_h.m
//===---------------------------------------------------------------------===//
C++ Template Instantiation benchmark:
http://users.rcn.com/abrahams/instantiation_speed/index.html
//===---------------------------------------------------------------------===//
TODO: File Manager Speedup:
We currently do a lot of stat'ing for files that don't exist, particularly
when lots of -I paths exist (e.g. see the <iostream> example, check for
failures in stat in FileManager::getFile). It would be far better to make
the following changes:
1. FileEntry contains a sys::Path instead of a std::string for Name.
2. sys::Path contains timestamp and size, lazily computed. Eliminate from
FileEntry.
3. File UIDs are created on request, not when files are opened.
These changes make it possible to efficiently have FileEntry objects for
files that exist on the file system, but have not been used yet.
Once this is done:
1. DirectoryEntry gets a boolean value "has read entries". When false, not
all entries in the directory are in the file mgr, when true, they are.
2. Instead of stat'ing the file in FileManager::getFile, check to see if
the dir has been read. If so, fail immediately, if not, read the dir,
then retry.
3. Reading the dir uses the getdirentries syscall, creating a FileEntry
for all files found.
//===---------------------------------------------------------------------===//
// Specifying targets: -triple and -arch
//===---------------------------------------------------------------------===//
The clang supports "-triple" and "-arch" options. At most one -triple and one
-arch option may be specified. Both are optional.
The "selection of target" behavior is defined as follows:
(1) If the user does not specify -triple, we default to the host triple.
(2) If the user specifies a -arch, that overrides the arch in the host or
specified triple.
//===---------------------------------------------------------------------===//
verifyInputConstraint and verifyOutputConstraint should not return bool.
Instead we should return something like:
enum VerifyConstraintResult {
Valid,
// Output only
OutputOperandConstraintLacksEqualsCharacter,
MatchingConstraintNotValidInOutputOperand,
// Input only
InputOperandConstraintContainsEqualsCharacter,
MatchingConstraintReferencesInvalidOperandNumber,
// Both
PercentConstraintUsedWithLastOperand
};
//===---------------------------------------------------------------------===//
Blocks should not capture variables that are only used in dead code.
The rule that we came up with is that blocks are required to capture
variables if they're referenced in evaluated code, even if that code
doesn't actually rely on the value of the captured variable.
For example, this requires a capture:
(void) var;
But this does not:
if (false) puts(var);
Summary of <rdar://problem/9851835>: if we implement this, we should
warn about non-POD variables that are referenced but not captured, but
only if the non-reachability is not due to macro or template
metaprogramming.
//===---------------------------------------------------------------------===//
We can still apply a modified version of the constructor/destructor
delegation optimization in cases of virtual inheritance where:
- there is no function-try-block,
- the constructor signature is not variadic, and
- the parameter variables can safely be copied and repassed
to the base constructor because either
- they have not had their addresses taken by the vbase initializers or
- they were passed indirectly.
//===---------------------------------------------------------------------===//