Patchwork [wwwdoc] : Add additional new features in 4.6

login
register
mail settings
Submitter Gerald Pfeifer
Date Jan. 3, 2011, 9:28 a.m.
Message ID <alpine.LNX.2.00.1101031022140.4999@gerinyyl.fvgr>
Download mbox | patch
Permalink /patch/77241/
State New
Headers show

Comments

Gerald Pfeifer - Jan. 3, 2011, 9:28 a.m.
On Sun, 2 Jan 2011, Kai Tietz wrote:
> This patch adds some new features of 4.6.

in that section.)  Just to make sure, the change per se is fine; I
just think it might be "Support for", here and in other parts of the
patch.

+    <li>New function attribute <code>callee_pop_aggregate</code> allows
+    to specify if caller, or callee is responsible to pop aggregate
+    return pointer value from stack.</li>

Omit the comma before "or" and add "the" before "aggregate" and "stack".

+<h3 id="windows">Windows x86/x86_64</h3>
+  <ul>
+    <li>Initial support of decimal floating point.</li>

"for" (also in several further items)

+    <li>Support of hot-patchable function prologue via attribute
+    <code>ms_hook_prologue> also for x86_64 architecture.</li>

So, we have this for x86 (32-bit already)?  Then we could say "for
x86-64 in addition to 32-bit x86" to make that more clear.

+    <li>Improvment of stack-probing and stack-allocation mechanism.</li>

"Improvements"
"mechanisms"

+    <li>Support of push/pop-macro pragma as preprocessor command.</li>

Can you provide a concrete code example?

> Ok for commit?

Yes, modulo the changes above -- and please omit those unrelated hunks
that only change whitespace in other parts of the page.

Thanks!

Gerald
Kai Tietz - Jan. 3, 2011, 10:11 a.m.
2011/1/3 Gerald Pfeifer <gerald@pfeifer.com>:
> On Sun, 2 Jan 2011, Kai Tietz wrote:
>> This patch adds some new features of 4.6.
>
> Index: wwwdocs/htdocs/gcc-4.6/changes.html
> ===================================================================
> --- wwwdocs.orig/htdocs/gcc-4.6/changes.html    2011-01-02 22:36:57.000000000 +0100
> +++ wwwdocs/htdocs/gcc-4.6/changes.html 2011-01-02 23:10:07.862964200 +0100
> @@ -180,6 +180,11 @@
>       calls to functions that return to the current unit only via returning
>       or exception handling.  This is the case for most library functions
>       that have no callbacks.</li>
> +    <li>Support of new type <code>__int128</code> for targets having
> +    wide enough machine-mode support.</li>
>
> So, this is beyond just x86?  (We already have an item

Yes, this feature depends on TImode >= 128-bit support and has nothing
to do with IA. It isn't related to the __float128 feature, as __int128
is an integer scalar ISO-C extension type and not a floating point
one.

Thanks for review. I will adjust text later this evening (I don't have
access from office to wwwdoc CVS).

Regards,
Kai
Gerald Pfeifer - Jan. 3, 2011, 10:40 a.m.
On Mon, 3 Jan 2011, Kai Tietz wrote:
> Yes, this feature depends on TImode >= 128-bit support and has nothing
> to do with IA. It isn't related to the __float128 feature, as __int128
> is an integer scalar ISO-C extension type and not a floating point
> one.

Ahem, of course __float128 and __int128 are _slightly_ different data
types.  How embarrassing.  /me crawls back into my cave now.

Gerald

Patch

Index: wwwdocs/htdocs/gcc-4.6/changes.html
===================================================================
--- wwwdocs.orig/htdocs/gcc-4.6/changes.html	2011-01-02 22:36:57.000000000 +0100
+++ wwwdocs/htdocs/gcc-4.6/changes.html	2011-01-02 23:10:07.862964200 +0100
@@ -180,6 +180,11 @@ 
       calls to functions that return to the current unit only via returning
       or exception handling.  This is the case for most library functions
       that have no callbacks.</li>
+    <li>Support of new type <code>__int128</code> for targets having
+    wide enough machine-mode support.</li>

So, this is beyond just x86?  (We already have an item 

  "Darwin, FreeBSD, MinGW and Cygwin now all support 
  <code>__float128</code> on 32-bit x86 targets."