Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 1124, Thu Mar 7 19:49:22 2002 UTC	revision 1138, Wed Mar 13 03:03:50 2002 UTC
#	Line 11	Line 11
11	Date: yyyy/mm/dd	Date: yyyy/mm/dd
12	Tag: <post-commit CVS tag>	Tag: <post-commit CVS tag>
13	Description:	Description:
14		----------------------------------------------------------------------
15		Name: Matthias Blume
16		Date: 2002/03/12 22:03:00 EST
17		Tag: blume-20020312-url
18		Description:
19
20		Forgot to update BOOT and srcarchiveurl.
21
22		----------------------------------------------------------------------
23		Name: Matthias Blume
24		Date: 2002/03/12 17:30:00 EST
25		Tag: blume-20020312-version110392
26		Description:
27
28		Yet another version number bump (because of small changes to the
29		binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
30
31		Changes:
32
33		The new pid generation scheme described a few weeks ago was overly
34		complicated.  I implemented a new mechanism that is simpler and
35		provides a bit more "stability":  Once CM has seen a compilation
36		unit, it keeps its identity constant (as long as you do not delete
37		those crucial CM/GUID/* files).  This means that when you change
38		and interface, compiler, then go back to the old interface, and
39		compile again, you arrive at the original pid.
40
41		There now also is a mechanism that instructs CM to use the plain
42		environment hash as a module's pid (effectively making its GUID
43		the empty string).  For this, "noguid" must be specified as an
44		option to the .sml file in question within its .cm file.
45		This is most useful for code that is being generated by tools such
46		as ml-nlffigen (because during development programmers tend to
47		erase the tool's entire output directory tree including CM's cached
48		GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
49		revert to the old, broken behavior of SML/NJ, but in specific cases
50		where there is no danger of interface confusion, its use is ok
51		(I think).
52
53		ml-nlffigen by default generates "noguid" annotations.  They can be
54		turned off by specifying -guid in its command line.
55
56		----------------------------------------------------------------------
57		Name: Lal George
58		Date: 2002/03/12 12 14:42:36 EST
59		Tag: george-20020312-frequency-computation
60		Description:
61
62		Integrated jump chaining and static block frequency into the
63		compiler. More details and numbers later.
64
65		----------------------------------------------------------------------
66		Name: Lal George
67		Date: 2002/03/11 11 22:38:53 EST
68		Tag: george-20020311-jump-chain-elim
69		Description:
70
71		Tested the jump chain elimination on all architectures (except the
72		hppa).  This is on by default right now and is profitable for the
73		alpha and x86, however, it may not be profitable for the sparc and ppc
74		when compiling the compiler.
75
76		The gc test will typically jump to a label at the end of the cluster,
77		where there is another jump to an external cluster containing the actual
78		code to invoke gc. This is to allow factoring of common gc invocation
79		sequences. That is to say, we generate:
80
81		f:
82		testgc
83		ja   L1      % jump if above to L1
84
85		L1:
86		jmp L2
87
88
89		After jump chain elimination the 'ja L1' instructions is converted to
90		'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
91		up being implemented in their long form (if L2 is far away) using:
92
93		jbe     L3      % jump if below or equal to L3
94		jmp     L2
95		L3:
96		...
97
98
99		For large compilation units L2  may be far away.
100
101
102		----------------------------------------------------------------------
103		Name: Matthias Blume
104		Date: 2002/03/11 13:30:00 EST
105		Tag: blume-20020311-mltreeeval
106		Description:
107
108		A functor parameter was missing.
109
110		----------------------------------------------------------------------
111		Name: Allen Leung
112		Date: 2002/03/11 10:30:00 EST
113		Tag: leunga-20020310-runtime-string0
114		Description:
115
116		The representation of the empty string now points to a
117		legal null terminated C string instead of unit.  It is now possible
118		to convert an ML string into C string with InlineT.CharVector.getData.
119		This compiles into one single machine instruction.
120
121		----------------------------------------------------------------------
122		Name: Allen Leung
123		Date: 2002/03/10 23:55:00 EST
124		Tag: leunga-20020310-x86-call
125		Description:
126
127		Added machine generation for CALL instruction (relative displacement mode)
128
129		----------------------------------------------------------------------
130		Name: Matthias Blume
131		Date: 2002/03/08 16:05:00
132		Tag: blume-20020308-entrypoints
133		Description:
134
135		Version number bumped to 110.39.1.  NEW BOOTFILES!
136
137		Entrypoints: non-zero offset into a code object where execution should begin.
138
139		- Added the notion of an entrypoint to CodeObj.
140		- Added reading/writing of entrypoint info to Binfile.
141		- Made runtime system bootloader aware of entrypoints.
142		- Use the address of the label of the first function given to mlriscGen
143		as the entrypoint.  This address is currently always 0, but it will
144		not be 0 once we turn on block placement.
145		- Removed the linkage cluster code (which was The Other Way(tm) of dealing
146		with entry points) from mlriscGen.
147
148		----------------------------------------------------------------------
149		Name: Allen Leung
150		Date: 2002/03/07 20:45:00 EST
151		Tag: leunga-20020307-x86-cmov
152		Description:
153
154		Bug fixes for CMOVcc on x86.
155
156		1. Added machine code generation for CMOVcc
157		2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
158		3. CMOVcc cannot have an immediate operand as argument.
159
160		----------------------------------------------------------------------
161		Name: Matthias Blume
162		Date: 2002/03/07 16:15:00 EST
163		Tag: blume-20020307-controls
164		Description:
165
166		This is a very large but mostly boring patch which makes (almost)
167		every tuneable compiler knob (i.e., pretty much everything under
168		Control.* plus a few other things) configurable via both the command
169		line and environment variables in the style CM did its configuration
170		until now.
171
172		Try starting sml with '-h' (or, if you are brave, '-H')
173
174		To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
175		implements the underlying generic mechanism.
176
177		The interface to some of the existing such facilities has changed somewhat.
178		For example, the MLRiscControl module now provides mkFoo instead of getFoo.
179		(The getFoo interface is still there for backward-compatibility, but its
180		use is deprecated.)
181
182		The ml-build script passes -Cxxx=yyy command-line arguments through so
183		that one can now twiddle the compiler settings when using this "batch"
184		compiler.
185
186		TODO items:
187
188		We should go through and throw out all controls that are no longer
189		connected to anything.  Moreover, we should go through and provide
190		meaningful (and correct!) documentation strings for those controls
191		that still are connected.
192
193		Currently, multiple calls to Controls.new are accepted (only the first
194		has any effect).  Eventually we should make sure that every control
195		is being made (via Controls.new) exactly once.  Future access can then
196		be done using Controls.acc.
197
198		Finally, it would probably be a good idea to use the getter-setter
199		interface to controls rather than ref cells.  For the time being, both
200		styles are provided by the Controls module, but getter-setter pairs are
201		better if thread-safety is of any concern because they can be wrapped.
202
203		*****************************************
204
205		One bug fix: The function blockPlacement in three of the MLRISC
206		backpatch files used to be hard-wired to one of two possibilities at
207		link time (according to the value of the placementFlag).  But (I
208		think) it should rather sense the flag every time.
209
210		*****************************************
211
212		Other assorted changes (by other people who did not supply a HISTORY entry):
213
214		1. the cross-module inliner now works much better (Monnier)
215		2. representation of weights, frequencies, and probabilities in MLRISC
216		changed in preparation of using those for weighted block placement
217		(Reppy, George)
218
219	----------------------------------------------------------------------	----------------------------------------------------------------------
220	Name: Lal George	Name: Lal George

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.1124   changed lines   Added in v.1138

---

root@smlnj-gforge.cs.uchicago.edu	ViewVC Help
Powered by ViewVC 1.0.0