sbcl Log


Commit Date  
[c06b8c] (21.1 kB) by Christophe Rhodes Christophe Rhodes

support for DragonFly BSD

x86(-64) only for now, and threads are currently non-functional.
Closes lp#1292845, thanks to Vasily Postnicov.

2014-06-20 10:48:35 View
Download
[4ee98d] (21.0 kB) by Alastair Bridgewater Alastair Bridgewater

Dummy in basics for ARM port.

* Added commentary for :ARM feature to base-target-features.

* Altered feature sanity-checker to know about ARM and believe
that it is a cheneygc target (can and probably will be changed
later, but a cheneygc bringup should be easier than a gencgc
bringup).

* Added an :ARM target mapping for source files.

* Added dummy files for each target-specific source file in the
build process.

2014-05-10 02:22:49 View
Download
[60deeb] (20.9 kB) by Paul Khuong Paul Khuong

Simpler word-sized variable right shifts on x86 and x86-64

* Known negative shifts are converted to another function that
only handles machine-friendly right shifts.

* The transforms and VOPs are conditionalised on ash-right-vops,
so other platforms aren't penalised.

* The new transforms trigger a lot of notes; this is suboptimal,
and one test had to be adjusted.

2013-05-22 05:13:22 View
Download
[b38f10] (20.8 kB) by Paul Khuong Paul Khuong

Front end infrastructure for short vector SIMD packs

* new feature, sb-simd-pack.

* define a new IR1 type for SIMD packs:
- (SB!KERNEL:SIMD-PACK [eltype]), where [eltype] is a subtype
of the plaform-specific SIMD element type universe, or * (default),
the union of all these possibilities;
- Element types are always upgraded to the platform's element type
(small) universe, so we can easily manipulate unions of SIMD-PACK
types by working in terms of the element types.

* immediately specify the universe of SIMD pack element types
(sb!kernel:*simd-pack-element-types*) for x86-64, to ensure
#!+sb-simd-pack buildability.

* declare basic functions to create/manipulate SIMD packs:
- simd-pack-p is the basic type predicate;
- %simd-pack-tag returns a fixnum tag associated with each SIMD-PACK;
currently, we suppose it only encodes the element type, as the
position of the element type in *simd-pack-element-types*;
- %make-simd-pack creates a 128-bit SIMD pack from a tag and two
64 bit integers;
- %make-simd-pack-double creates an appropriately-tagged pack from
two double floats;
- %make-simd-pack-single creates a tagged pack from four single
floats;
- %make-simd-pack-ub{32,64} creates a tagged pack from four 32 bit
or two 64 bit integers;
- %simd-pack-{low,high} returns the low/high integer half of a
128 bit pack;
- %simd-pack-ub{32,64}s returns the four integer quarters or two
integer halves of a 128 bit pack;
- %simd-pack-singles returns the four singles in a 128 bit pack;
- %simd-pack-doubles returns the two doubles in a 128 bit pack.

Alexander Gavrilov kept a branch alive for the last couple years. The
creation/manipulation primitives are largely taken from that branch,
or informed by the branch's usage.

2013-05-21 19:10:50 View
Download
[098f53] (20.7 kB) by Nikodemus Siivola Nikodemus Siivola

oops, it's :package-local-nicknams, not :local-nicknames

Also mention it in the manual.

...just couldn't get it right the first time.

2013-02-01 16:40:23 View
Download
[b0b221] (20.7 kB) by Nikodemus Siivola Nikodemus Siivola

package local nicknames

Example terminal session using Linedit:

* (defpackage :foo (:use :cl) (:local-nicknames (:sb :sb-ext)))

#<PACKAGE "FOO">
* (in-package :foo)

#<PACKAGE "FOO">
* (sb:posix-
sb:posix-environ sb:posix-getenv
* (sb:posix-getenv "USER")

"nikodemus"

API:

function PACKAGE-LOCAL-NICKNAMES package
function PACKAGE-LOCALLY-NICKNAMED-BY package
function ADD-PACKAGE-LOCAL-NICKNAME nick global &optional package
function REMOVE-PACKAGE-LOCAL-NICKNAME old-nick &optional package

DEFPACKAGE option: (:local-nicknames {(local-nick global-name)}*)

:PACKAGE-LOCAL-NICKNAMES in *FEATURES*

Design issues and considerations:

* "CL", "COMMON-LISP", and "KEYWORD" signal a continuable error
when used as local nicknames. I think this is good for sanity,
but not strictly required. Because of the way (find-package
:keyword) is idiomatically used to guarantee print/read
consistency across packages, I think it at least should be
protected.

* To preserve read/print consistency, we use package local nicknames
as prefixes when printing.

* The hook into FIND-PACKAGE is invisible, but built on top of
FIND-PACKAGE-USING-PACKAGE -- undocumented and unexported, but
waiting to be turned into something interesting by Christophe.

* Local nicknames are protected by package locks.

* If you want to bypass nicknames, you need to first get into
a known package without nicknames. There could be an explicit
way as well, but not sure if that's needed or a good idea.

Random crap mixed in:

Re-order DEFPACKAGE option docs in rough order of usefulness.

2013-02-01 16:22:39 View
Download
[9644b3] (20.5 kB) by David Lichteblau David Lichteblau

In x86 arch_os_get_current_thread(), do not load from %fs

Simplify arch_os_get_current_thread() such that %fs is not being
loaded directly even on platforms which preserve it reliably (in
particular, Linux/x86 and Solaris/x86).

Aside from a code simplification and reduction of cross-platform
differences, this change is an improvement in the following way:

If arch_os_get_current_thread() gets called in a non-Lisp pthread,
it now returns NULL cleanly even on the platforms mentioned above.

On the other platforms, keep the support for restoring %fs, but
refactor to remove code duplication with arch_os_thread_init().
Coalesce the two differently-spelled Lisp features for this case
into one.

2012-12-17 13:42:07 View
Download
[266ccb] (20.5 kB) by David Lichteblau David Lichteblau

Add a safepoint-based mechanism to avoid SIGALRM for the TIMER facility

- Retrofits the signal-free timer thread for Windows (thanks to
Anton Kovalenko) to POSIXy platforms.

- Provide os_* functions in the C runtime which simulate the win32
API for waitable timers.

Currently supported on Linux (timerfd), FreeBSD (kqueue), and SunOS
(completion ports). A tentative (untested) implementation is
provided for Darwin's kqueue.

2012-09-19 15:10:28 View
Download
[ebb604] (20.2 kB) by David Lichteblau David Lichteblau

Add odxprint, a replacement for FSHOW which can be configured at run-time

- A new macro odxprint(flag, "fmt", ...) performs the equivalent of
a printf("fmt", ...), but only if `flag' has been enabled at
run-time.

- Environment variables can be used to set flags, using either
SBCL_DYNDEBUG="flag1 flag2 flag3" syntax, or
SBCL_DYNDEBUG__FLAG1="nonempty string" syntax.

- Lisp feature SB-QSHOW enables support for odxprint-based FSHOW.
(Users who prefer to edit runtime.h to enable QSHOW can still do
so...) SB-QSHOW is enabled by default on Windows, where the
odxprint mechanism was first used.

- Implement FSHOW, FSHOW_SIGNAL on top of odxprint. Corresponding
flags are called fshow, fshow_signal.

- For gencgc_verbose, support a flag of the same name, since it is
conditional on QSHOW (inspite of not being implemented on top of
FSHOW).

- Does not yet support odxprint features specific to Windows debugger
integration; output is currently directed to stderr unconditionally.

This commit backports Anton Kovalenko's Windows-specific odxprint to
POSIX and integrates it with FSHOW.

2012-09-11 10:14:50 View
Download
[8cef5f] (20.0 kB) by David Lichteblau David Lichteblau

Mention sb-safepoint, sb-thruption in base-target-features.lisp-expr

2012-09-10 16:32:02 View
Download
[d6f967] (19.3 kB) by Nikodemus Siivola Nikodemus Siivola

killing lutexes, adding timeouts

* Remove all lutex-specific code from the system.
** Use SB-FUTEX for futex-capable platforms, and plain SB-THREAD
otherwise.
** Make non-futex mutexes unfair spinlocks for now, using WAIT-FOR to
provide timeouts and backoff.
** Build non-futex condition variables on top of a queue and WAIT-FOR.

Performance implications: SB-FUTEX builds should perform pretty much the
same, or improve a bit. Threaded non-futex builds are affected as follows:

1. Threads idling on semaphores or condition variables aren't quite as
cheap. Just how costly depends on the OS. On Darwin 1000 idle threads
can chew up a bit over 50% CPU. I will try to address this later.

2. Contested locking around operations that take considerably longer
than a single timeslice suffers mild degradation.

3. Contested locking around operations that don't take long is an order
of magnitude performant.

4. Highly active semaphores perform much better. (Follows from #3.)

* GRAB-MUTEX gets timeout support on all platforms.

* CONDITION-WAIT gets timeout support.

* Disable a bunch of prone-to-hang thread tests on Darwin. (All of them
were already prone to hang prior to this commit.)

* Enable a bunch tests that now /pass/ on Darwin. \o/ This doesn't mean that
the threaded Darwin is fully expected to pass all tests yet, but let's say
it's more likely to do so.

...but still not robust enough to enable threads on Darwin by default.

* GET-MUTEX/GRAB-MUTEX get refactored into two main parts: %TRY-MUTEX and
%WAIT-ON-MUTEX, which are also used directly from CONDITION-WAIT where
appropriate.

2011-11-09 23:00:48 View
Download
[797692] (19.5 kB) by Paul Khuong Paul Khuong

Optional support for zlib-based in-memory deflate/inflate for core files

* As this is based on zlib, only add the dependency when
:SB-CORE-COMPRESSION is enabled as a build-time feature. On x86-64,
compressed cores take about 1/4 the space, but start up in a few
tenths of a second.

Unlike gzexe'd executables, compressed images work without writing
to /tmp.

If :SB-CORE-COMPRESSION is enabled, trigger compression with the
:COMPRESSION argument to SAVE-LISP-AND-DIE.

* Also add a NEWS entry for the literal complex-single-float bugfix

2011-08-28 03:23:03 View
Download
[3254e1] (19.3 kB) by Paul Khuong Paul Khuong

New function: SB!KERNEL:%MULTIPLY-HIGH

* Does the same thing as only returning the first value of %MULTIPLY,
only better on some platforms.

* Implemented vas VOPs on x86, x86-64 and PPC. The PPC code sequence
is fully untested, and merely looks correct.

* VOPs for fixnum first argument are included, but will only be used
when the result is forcibly marked as fixnum, e.g., with TRULY-THE.
Questionnable, but I'd rather err on the side of straightforwardness
rather than put even more pressure on representation selection.

* Use it in the division-by-multiplication transform for unsigned
TRUNCATE by constant.

2011-08-14 20:49:27 View
Download
[65b5ab] (19.1 kB) by Alastair Bridgewater Alastair Bridgewater

"1.0.41.2": threads: Add memory-barrier framework.

* New file, src/code/barrier.lisp, containing the baseline
implementation.

* Added the barrier functions to the compiler function database.

* Export the interface macro, BARRIER, from SB!THREAD and the
underlying barrier functions from SB!VM.

* Document a new architecture-dependent build-time feature,
MEMORY-BARRIER-VOPS, for controlling the behavior and inlining of
the interpreter stubs for the barrier functions.

2010-08-04 17:58:15 View
Download
[11b5ac] (18.9 kB) by Alastair Bridgewater Alastair Bridgewater

1.0.36.9: UD2-BREAKPOINTS feature for x86oid systems

* Add new feature UD2-BREAKPOINTS, enabled by default only on x86oid
darwin targets.

* Use said feature instead of DARWIN for breakpoint trap selection.

* Make breakpoints work when using UD2-BREAKPOINTS (tested on x86 and
x86-64 linux).

* This patch brought to you by lp#309067, which remains valid for
three reasons: First, the test case is still disabled. Second, this
only fixes for x86oids, not for PPC. And third, I didn't actually test
this on a darwin system.

2010-03-01 13:09:00 View
Download
[1baab0] (18.7 kB) by Nikodemus Siivola Nikodemus Siivola

1.0.33.26: enabled threads by default on x86[-64] Linux

I'll let FreeBSD folks make the judgement if threads should be
default there as well.

Also: update INSTALL documentation regarding *FEATURES* a bit, and
make the documentation clear about availability of threads on
different builds.

2009-12-18 14:26:33 View
Download
[49e840] (18.9 kB) by trittweiler trittweiler

1.0.32.29: Add build flag :sb-xref-for-internals.

Enabling :sb-xref-for-internals in customize-target-features.lisp,
will make Sbcl collect Xref data about itself during the build. This
increases the core size drastically by about 5-6mb, but it's useful
for SBCL developers because they can now use M-? (slime-edit-uses) to
get a list of call/expansion/reference sites for internal stuff.

It may be interesting to Lisp advocacy who can now show off with
finding the use sites of standardized functions like CONS, etc. :-)

Additionally -- regardless of :sb-xref-for-internals --, we now also
collect xref data for keywords because they're "fine" names for
functions and macros, and I know of people who use MACROLET on
keywords for their DSLs.

2009-11-12 15:10:04 View
Download
[2230ea] (18.6 kB) by Paul Khuong Paul Khuong

1.0.29.54: Inline unboxed constants on x86[-64]

* New build-time feature: inline-constants, which specifies that SB!C
and SB!VM implement a protocol described in base-target-features.lisp-expr.
Backends implementing that feature are able to load constants from code
components, in a section that follows the actual executable code.

* Implement the protocol on x86 and x86-64, and use it for float constants,
and, on x86-64 only, mid-sized (> 2^(29-32), but still machine-sized)
integers.

* Use the new feature in integer and float arithmetic VOPs.

* Adjust a few test cases to take newly consing situations into account.

* Clean-up:
- New build-time feature: float-eql-vops, which disable rewriting EQL
of single and double floats in terms of foo-float*-bits.
- Fix a typo (unused variable lookup) in TWO-ARG-+/-

2009-06-28 21:37:05 View
Download
[a157ed] (16.9 kB) by Paul Khuong Paul Khuong

1.0.29.44: Complex float improvements

* On all platforms:
- Slightly more stable complex-complex float (double and single)
division;
- New transform for real-complex division;
- complex-real and real-complex float addition and subtraction
behave as though the real was first upgraded to a complex, thus
losing the sign of any imaginary zero.

* On x86-64
- Complexes floats are represented packed in a single SSE register;
- VOPs for all four arithmetic operations, complex-complex, but also
complex-real and real-complex, except for complex-complex and
real-complex division;
- VOPs for =, negate and conjugate of complexes (complex-real and
complex-complex);
- VOPs for EQL of floats (real and complexes).
- Full register moves for float values in SSE registers should also
speed scalar operations up.

2009-06-25 15:37:05 View
Download
[dcd860] (16.3 kB) by Nikodemus Siivola Nikodemus Siivola

1.0.24.12: adding and fixing the HPUX/HPPA build target

* Patch by Larry Valkama.

2009-01-03 15:50:46 View
Download
[eb01bd] (16.2 kB) by Nikodemus Siivola Nikodemus Siivola

1.0.16.38: document :CYCLE-COUNTER feature in base-target-features.lisp-expr

* Weaseling out from renaming it to :SB-CYCLE-COUNTER, and deferring that
a bit: seems better to take care of all similar features that are really
just built-time conveniences automatically inferred at the same time.

2008-05-19 13:47:17 View
Download
[f318d0] (15.8 kB) by Juho Snellman Juho Snellman

1.0.10.14: remove locking and gc inhibition from hash-tables, power of 2 sizes

This commit removes a bunch of bottlenecks from the hash-table
implementation. It speeds up GETHASH, (SETF GETHASH) and
REMHASH by a factor of 2-4x (on platforms with a real
WITH-PINNED-OBJECTS) depending on the operation. On the flip
side, no automatic locking is done on tables any more, so
multi-threaded applications must do their own locking. (The
locking done by SBCL was always just an implementation detail,
not a part of the external interface). By popular demand it's
also still safe to have multiple readers on the same table
without locking.

Originally GCs were inhibited during most hash-table
operations for two reasons. To prevent the GC from rehashing a
table while a Lisp-side operation is going on, and to prevent
the GC from moving the key after the hash-value has been
calculated.

More recently, most hash-tables operations have acquired a
lock on the table in order to prevent two concurrent writers
from corrupting the chains. While it's never been the intent
for the standard data structures to be automatically
thread-safe in SBCL, this locking had to be done since corrupt
tables could lead to infinite GC loops.

Both the locking and the without-gcing are expensive
operations relative to the total cost of a hash-table lookup.
This commit removes both the gc inhibition and the locks.
Additionally we switch to power of two table size, which
allows calculating a cheaper hash -> bucket with cheaper
operations than MOD.

* The GC no longer does the rehashing itself, but just marks
the hash-table as needing a rehash, which will then be done
Lisp-side when the table is next accessed. While it's
possible to find cases where the former behaviour has better
performance, they're very contrived.
* The hash-table operations that work on the chains now check
for loops in the chains, and signal an error if one is found.
* The hash-table operations now pin the key before calculating
the hash value (needed for EQ-based hash functions).
* Add a GC epoch value that GETHASH can use to check whether
a GC happened during the lookup. This is needed since another
thread calling GETHASH on the same table might have caused it
to be rehashed.
* Kill the old MUST-REHASH vector header, and replace it with a
slot in the HASH-TABLE structure. The overloading of the header
caused missed rehashings when both the GC and %%PUTHASH modified
it at the same time.
* Switch to power of two table sizes, with a slightly more complex
hash value -> bucket calculation than just taking the low bits,
which in many cases have a very skewed distribution for the existing
SBCL hash functions. Still a lot faster than using MOD.
* Leave in locking and GC inhibition during rehashing (needed to
allow multiple readers to coexist) and for weak hash-tables
(they need some GC support, and the code is much simpler when
all of the logic is in the GC instead of interleaved in the GC and
Lisp-side). Neither of these cases is performance critical.

2007-09-30 23:18:50 View
Download
[6d3a96] (15.5 kB) by Thiemo Seufer Thiemo Seufer

1.0.9.26: Update comment about little-endian MIPS.

2007-09-02 03:04:32 View
Download
[b822fd] (15.5 kB) by Nikodemus Siivola Nikodemus Siivola

1.0.9.6: remove :UNIX from *FEATURES* on Windows

Thanks to Luis Oliveira.

2007-08-29 14:51:55 View
Download
[bfb19d] (15.7 kB) by Nikodemus Siivola Nikodemus Siivola

1.0.7.19: SB-EXT:COMPARE-AND-SWAP

* New macro SB-EXT:COMPARE-AND-SWAP provides a supported interface to
compare-and-swap functionality.

* New info-type :FUNCTION :STRUCTURE-ACCESSOR allows us to map from
defstruct slot-accessor names to defstruct descriptions.

* Add :CAS-TRANS slot keyword to DEFINE-PRIMITIVE object, and the
compiler machinery needed to support compare and swap on primitive
object slots.

* New VOPs COMPARE-AND-SWAP-SLOT and %COMPARE-AND-SWAP-SYMBOL-VALUE.

* Delete now unnecessary DEFINE-STRUCTURE-SLOT-COMPARE-AND-SWAP.

* Use a consistent %COMPARE-AND-SWAP-FOO naming scheme for CAS
functions.

* Tests.

Tested on x86/Linux & x86/Darwin, x86-64/Darwi, and PPC/Darwin.

2007-07-15 22:28:12 View
Download
Older >