mirrors/emacs
1
Fork 0
mirror of git://git.sv.gnu.org/emacs.git synced 2026-03-26 08:41:47 -07:00
Code Issues Projects Releases Wiki Activity
4884 commits 248 branches 194 tags 1.1 GiB
Commit graph

75 commits

Author SHA1 Message Date
Gareth Rees
0ffc9ec788 Can't use checkd_nosig(tree, ...) because treeempty is null. 
arena->enabledMessageTypes might be NULL.

Copied from Perforce
 Change: 185271
 ServerID: perforce.ravenbrook.com
 2014-04-07 10:14:40 +01:00
Gareth Rees
bcea6812fc Improve control over checking: 
1. Where Type is a pointer type with a signature, replace CHECKL(TypeCheck(val)) with CHECKD(Type, val).
2. Where Type is a pointer type with no signature, replace CHECKL(TypeCheck(val)) with CHECKD_NOSIG(Type, val).
3. Where Type is a pointer type with a signature, but the structure is not visible at point of checking, replace CHECKL(TypeCheck(val)) with CHECKD_NOSIG(Type, val). Reference <design/check/#.hidden-type>
4. Make BTCheck extern and use it where possible.
5. Replace AVER(TypeCheck(val)) with AVERT(Type, val).

Copied from Perforce
 Change: 185263
 ServerID: perforce.ravenbrook.com
 2014-04-06 22:51:05 +01:00
Richard Brooksby
6cfc226ebd Implementing top-down (last fit) allocation in zones. 
Copied from Perforce
 Change: 184771
 ServerID: perforce.ravenbrook.com
 2014-03-13 12:06:28 +00:00
Richard Brooksby
2a7fd3c326 Incomplete response to code review. see <https://info.ravenbrook.com/mail/2014/03/07/11-21-03/0/>. 
Copied from Perforce
 Change: 184735
 ServerID: perforce.ravenbrook.com
 2014-03-11 16:39:11 +00:00
Richard Brooksby
098eb9e6d7 Catch-up merge from master. 
Copied from Perforce
 Change: 184732
 ServerID: perforce.ravenbrook.com
 2014-03-11 16:30:47 +00:00
Richard Brooksby
4a27ce10f0 Catch-up merge from branch/2014-02-22/splay-tune to pick up code review changes. 
Copied from Perforce
 Change: 184727
 ServerID: perforce.ravenbrook.com
 2014-03-11 16:07:19 +00:00
Richard Brooksby
0cb5069ea8 Responding to code review <https://info.ravenbrook.com/mail/2014/03/07/22-40-36/0/>. 
Using TREE_ELT where appropriate.
Using AVERT consistently on Tree, but see job003691.

Copied from Perforce
 Change: 184724
 ServerID: perforce.ravenbrook.com
 2014-03-11 15:46:51 +00:00
Richard Brooksby
519f208a46 Responding to code review. see <https://info.ravenbrook.com/mail/2014/03/04/12-02-27/0/>. 
Copied from Perforce
 Change: 184669
 ServerID: perforce.ravenbrook.com
 2014-03-07 15:00:54 +00:00
Richard Brooksby
0112bd03b6 Resolving questions left behind in fixmes. 
Copied from Perforce
 Change: 184609
 ServerID: perforce.ravenbrook.com
 2014-02-28 14:42:44 +00:00
Richard Brooksby
c0e96f9f3c Merging branch/2014-02-22/splay-tune to pick up splay tree improvements. 
Copied from Perforce
 Change: 184606
 ServerID: perforce.ravenbrook.com
 2014-02-28 14:27:40 +00:00
Richard Brooksby
a2aa56447c Fixing res/bool confusion found by “xc/debug/fbmtest 244032167” 
Copied from Perforce
 Change: 184595
 ServerID: perforce.ravenbrook.com
 2014-02-28 10:53:07 +00:00
Richard Brooksby
cd19bda000 Tidying up interfaces to cbsinit and cbsfindinzones. 
Documenting assumptions about CBS block allocation used by the arena.

Copied from Perforce
 Change: 184593
 ServerID: perforce.ravenbrook.com
 2014-02-28 10:30:32 +00:00
Richard Brooksby
0c16e904f5 Simplifying the interface to cbsfindinzones. 
Speeding up CBSFindInZones a bit.

Copied from Perforce
 Change: 184589
 ServerID: perforce.ravenbrook.com
 2014-02-28 01:25:18 +00:00
Richard Brooksby
fc720aff46 Eliminating complex zonedcbs adt and reintroducing a cbs with zone summaries. removing the default “high” segment preference, that caused the cbs trees to unbalance, and isn’t required outside of ep. considerable reduction in code and speed up, now at parity with master on test xc/release/gcbench -x 1234 amc 
Copied from Perforce
 Change: 184587
 ServerID: perforce.ravenbrook.com
 2014-02-28 00:35:05 +00:00
Richard Brooksby
9dff6e6d1d Searching the freecbs now respects the “high” preference, though not yet within blocks. 
Copied from Perforce
 Change: 184574
 ServerID: perforce.ravenbrook.com
 2014-02-26 19:29:05 +00:00
Richard Brooksby
a06ba218b6 Fixing cbsfindfirstinzones to return a res, because it can fail when it can’t allocate a cbs block, unlike cbsfindfirst. 
Fixing related corner case in arenaAllocFindInFreeCBS when it thought it couldn’t find a block when it was in fact running out of nodes.  This was revealed by “gcbench --npass 1 --arena-size 1M --seed 945632066 --niter 1 amc”.
Fixing draft ArenaFreeCBSDelete after exercising it.

Copied from Perforce
 Change: 184568
 ServerID: perforce.ravenbrook.com
 2014-02-26 17:40:39 +00:00
Richard Brooksby
cffd3e4834 Backing out zone set summary in the cbs. this isn't needed for speed now that there's a cbs per zone in the arena, and it will slow down all other cbss. 
Copied from Perforce
 Change: 184561
 ServerID: perforce.ravenbrook.com
 2014-02-26 14:24:34 +00:00
Gareth Rees
32d10cd02a Make mps_args_done obsolete. 
Copied from Perforce
 Change: 184543
 ServerID: perforce.ravenbrook.com
 2014-02-26 12:47:27 +00:00
Richard Brooksby
01ee2406aa Fixing bug where cbs overwrites its own blockpool pointer with garbage. 
Copied from Perforce
 Change: 184532
 ServerID: perforce.ravenbrook.com
 2014-02-25 19:19:37 +00:00
Richard Brooksby
30e05ec9be Permitting tree traversals to abort early. 
Copied from Perforce
 Change: 184511
 ServerID: perforce.ravenbrook.com
 2014-02-23 15:36:27 +00:00
Richard Brooksby
72dcf4aa74 Removing unnecessary case of splaying empty trees. 
Copied from Perforce
 Change: 184450
 ServerID: perforce.ravenbrook.com
 2014-02-23 11:52:21 +00:00
Richard Brooksby
a7451a424a Adding a signature to splay tree structure. miscellaneous documentation and naming clean-up. 
Copied from Perforce
 Change: 184448
 ServerID: perforce.ravenbrook.com
 2014-02-23 10:41:15 +00:00
Richard Brooksby
ee50a7703d Implementing pointer-reversing tree traversal that does not disturb the tree shape, unlike the splay tree traversal, which flattens the tree into a list. replacing cbsiterate with this, knocking about 25% off the runtime of the test case: xc/release/djbench -x 1234 --sshift 8 mvff 
Copied from Perforce
 Change: 184444
 ServerID: perforce.ravenbrook.com
 2014-02-23 10:41:15 +00:00
Richard Brooksby
2a60898a04 Eliminating redundant passing of tree keys by adding a method to extract the key from a tree node. 
Copied from Perforce
 Change: 184440
 ServerID: perforce.ravenbrook.com
 2014-02-23 10:41:15 +00:00
Richard Brooksby
8fccc81c82 Avoiding general compare when searching for first and last nodes in a tree. 
Copied from Perforce
 Change: 184435
 ServerID: perforce.ravenbrook.com
 2014-02-23 10:41:15 +00:00
Richard Brooksby
a97757482b Fixing abuses of res and resfail in the splay tree interface. 
Copied from Perforce
 Change: 184432
 ServerID: perforce.ravenbrook.com
 2014-02-23 10:41:15 +00:00
Richard Brooksby
0b3372d1d2 Distinguishing tree keys (treekey) from generic void *. 
Copied from Perforce
 Change: 184429
 ServerID: perforce.ravenbrook.com
 2014-02-23 10:41:15 +00:00
Richard Brooksby
07ece0b4b3 Unifying splaycomparemethod with treecompare. 
Copied from Perforce
 Change: 184428
 ServerID: perforce.ravenbrook.com
 2014-02-23 10:41:15 +00:00
Richard Brooksby
a0435161f2 Simplifying splay tree update prototype and removing special null case. 
Copied from Perforce
 Change: 184427
 ServerID: perforce.ravenbrook.com
 2014-02-23 10:41:15 +00:00
Richard Brooksby
8a7f2671f4 Replacing splaynode with tree generally. 
Copied from Perforce
 Change: 184426
 ServerID: perforce.ravenbrook.com
 2014-02-23 10:41:15 +00:00
Richard Brooksby
b5f1710b3b Replacing splaynodestruct with treestruct and basic tree navigation operations in the splay tree implementation. 
Copied from Perforce
 Change: 184425
 ServerID: perforce.ravenbrook.com
 2014-02-23 10:41:15 +00:00
Richard Brooksby
cedb0c1c9b Desmurfing the naming conventions in cbs. 
Copied from Perforce
 Change: 184424
 ServerID: perforce.ravenbrook.com
 2014-02-23 10:41:14 +00:00
Richard Brooksby
46e7234676 Added a pool argument to cbs creation so that cbss can share a pool. in particular, per-zone allocation cbss for the arena can share a single special block pool. 
Copied from Perforce
 Change: 184469
 ServerID: perforce.ravenbrook.com
 2014-02-18 13:53:56 +00:00
David Lovemore
bbd710d806 Add zone set to cbs nodes for faster zone set allocation. 
Copied from Perforce
 Change: 184299
 ServerID: perforce.ravenbrook.com
 2014-02-05 15:04:40 +00:00
Richard Brooksby
33a73ce392 Non-working preferential allocation within a block of the cbs. 
This doesn't work because the splay tree can't cope with homing in on a node (that appears big enough) then being told that it isn't suitable (because it doesn't have space in the right zone).  The splay tree summary condition must be precise, as stated in the design.  The splay tree isn't suited to this kind of search.

Copied from Perforce
 Change: 184073
 ServerID: perforce.ravenbrook.com
 2014-01-21 16:29:30 +00:00
Richard Brooksby
cf8738fd9f First draft of allocating using the freecbs. arenaalloc finds a free address range in the cbs and asks the arena class to mark it allocated, rather than the arena class doing the search. plenty still missing. no freeing. 
Copied from Perforce
 Change: 184050
 ServerID: perforce.ravenbrook.com
 2014-01-18 17:16:10 +00:00
Richard Brooksby
454ad555c3 Inlining the mfs block pool's structure into its parent mfs, to avoid allocating in the control pool when initialising a cbs. this will allow the cbs to be used before the control pool exists. 
Copied from Perforce
 Change: 184043
 ServerID: perforce.ravenbrook.com
 2014-01-17 15:48:12 +00:00
Gareth Rees
4232dc3125 Tidying up in anticipation of review. 
Copied from Perforce
 Change: 182524
 ServerID: perforce.ravenbrook.com
 2013-06-04 23:49:48 +01:00
Gareth Rees
5fb6436dd0 Cbsdelete() now returns the isolated contiguous range that was found, even if the requested deletion operation cannot be performed. (this is so that the caller can try deleting the whole block instead and manage the fragments using a fallback strategy.) 
CBSFindLargest() takes a size argument, so that the caller doesn't have to re-insert the found block if it wasn't large enough.

Copied from Perforce
 Change: 182431
 ServerID: perforce.ravenbrook.com
 2013-06-03 16:16:04 +01:00
Gareth Rees
7dc558b489 Rename cbstest.c to fbmtest.c (free block management test) and generalize it so that it tests both the cbs and freelist modules. 
Split FreelistFind into FreelistFindFirst and FreelistFindLast so that the interface exactly matches the CBS interface.
Fix errors in freelist.c: missing computation of size in FreelistBlockSetLimit; update linked list correct in coalesceRight case in FreelistInsert.
Avoid compiler warnings in freelist.c.
In cbs.c: use $U for Booleans ($B is for bit tables).

Copied from Perforce
 Change: 182397
 ServerID: perforce.ravenbrook.com
 2013-06-02 21:34:23 +01:00
Gareth Rees
71a2c572a0 Use range objects in the cbs interface instead of base, limit pairs. the idea is that freelist and cbs should offer similar interfaces so that the testing code can be shared. 
Copied from Perforce
 Change: 182364
 ServerID: perforce.ravenbrook.com
 2013-05-31 16:29:26 +01:00
Gareth Rees
e7559cf22f Rename the enumeration cbsfinddelete to finddelete in anticipation of it being shared between cbs and new freelist module. 
Since ABQDisposition enumeration only has two values, it's better to use Bool.

Copied from Perforce
 Change: 182347
 ServerID: perforce.ravenbrook.com
 2013-05-31 00:09:33 +01:00
Gareth Rees
fcc50ead4c Clarify the logic in cbsinsertintotree to match cbsdeletefromtree. 
Copied from Perforce
 Change: 182344
 ServerID: perforce.ravenbrook.com
 2013-05-30 20:09:01 +01:00
Gareth Rees
467c38a37a Separate block allocation from block insertion, so that cbsdeletefromtree no longer leaves the tree in an inconsistent state if allocation fails. 
Simplify the case logic in cbsDeleteFromTree to try to make it clearer.
Explain why cbsDeleteFromTree can't fail when it's called from cbsFindDeleteRange.

Copied from Perforce
 Change: 182342
 ServerID: perforce.ravenbrook.com
 2013-05-30 19:22:56 +01:00
Gareth Rees
46ca22c648 Fix review comments from rb in <https://info.ravenbrook.com/mail/2013/05/23/18-17-17/0/> 
Copied from Perforce
 Change: 182149
 ServerID: perforce.ravenbrook.com
 2013-05-24 00:48:19 +01:00
Richard Brooksby
98ba0e67a5 Deleting some obsolete comment text about callbacks. 
Renaming cbsBlockShrink to cbsBlockShrunk and cbsBlockGrow to cbsBlockGrew, since they don't *do* those things, just record that they were done.
Tidying up Find functions, where deletion of emergency code had left some strange double nestings.

Copied from Perforce
 Change: 182120
 ServerID: perforce.ravenbrook.com
 2013-05-23 19:05:25 +01:00
Gareth Rees
4aaef08a42 Remove unused variable oldsize. 
Copied from Perforce
 Change: 182111
 ServerID: perforce.ravenbrook.com
 2013-05-23 12:50:57 +01:00
Gareth Rees
0ff7dacbb8 Catch-up merge from master sources to branch/2013-05-17/emergency. 
Copied from Perforce
 Change: 182048
 ServerID: perforce.ravenbrook.com
 2013-05-21 18:31:50 +01:00
Gareth Rees
5b5f5b8050 Cbsdescribe provides more information about the cbs. 
Copied from Perforce
 Change: 182030
 ServerID: perforce.ravenbrook.com
 2013-05-21 13:25:17 +01:00
Gareth Rees
4b970759be Make the cbs module more abstract by removing cbsblock from the public interface. avoid re-entrancy problems by removing the callback interface. public interfaces like cbsiteratemethod now operate in terms of address ranges rather than cbsblocks. 
The functions CBSInsert, CBSDelete and CBSFind* now additionally return an "old" address range which gives the former base and limit of the block that has just been updated. This gives clients enough information to update their caches if need be.
Update CBS test and design accordingly.

Copied from Perforce
 Change: 182014
 ServerID: perforce.ravenbrook.com
 2013-05-20 20:45:26 +01:00