[PATCH] gcc-13: Add changelog for LoongArch.

[PATCH] gcc-13: Add changelog for LoongArch.

[Lulu Cheng]

---
 htdocs/gcc-13/changes.html | 39 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 39 insertions(+)

diff --git a/htdocs/gcc-13/changes.html b/htdocs/gcc-13/changes.html
index f3b9afed..c75e341b 100644
--- a/htdocs/gcc-13/changes.html
+++ b/htdocs/gcc-13/changes.html
@@ -563,6 +563,45 @@ a work-in-progress.</p>
   </li>
 </ul>
 
+<h3 id="loongarch">LoongArch</h3>
+  <ul>
+    <li>New features</li>
+      <ul>
+        <li>The new command-line option <code>-mexplicit-relocs</code> decides whether
+	    to use or not use the assembler relocation operator when dealing with
+	    symbolic addresses. <code>-mexplicit-relocs</code> is enabled by default
+	    when GCC is configured to use a compatible assembler.
+        </li>
+        <li>Avoid using the GOT to access external symbols when the new
+	    <code>-mdirect-extern-access</code> command-line option is specified.
+        </li>
+        <li>The <a href="https://gcc.gnu.org/onlinedocs/gcc/LoongArch-Variable-Attributes.html#LoongArch-Variable-Attributes"><code>model</code></a>
+	    variable attributes has been added.
+        </li>
+      </ul>
+    <li>Built-in functions support Improvements</li>
+      <ul>
+        <li>The <code>rint</code> and <code>copysign</code> mathematical builtins
+	    (and their float variants) are now inplemented as inline LoongArch intrinsics.
+        </li>
+        <li>The <code>lrint</code>, <code>logb</code>, <code>scalbln</code>,
+	    <code>scalbn</code> and <code>ldexp</code> mathematical builtins (and their
+	    float variants) are now inplemented as inline LoongArch intrinsics when using
+	    <code>-fno-math-errno</code>.
+        </li>
+        <li>The <code>lceil</code> and <code>lfloor</code> mathematical builtins
+	    (and their float variants) are now inplemented as inline LoongArch intrinsics
+	    when using <code>-ffp-int-builtin-inexact</code>.
+        </li>
+      </ul>
+    <li>Subprojects Support</li>
+      <ul>
+        <li><code>libvtv</code> now supports LoongArch architecture.</li>
+        <li><code>libitm</code> now supports LoongArch architecture.</li>
+	<li>LoongArch supports address sanitizers other than <code>hwasan</code> and <code>tsan</code>.</li>
+      </ul>
+  </ul>
+
 <!-- <h3 id="mips">MIPS</h3> -->
 
 <!-- <h3 id="mep">MeP</h3> -->
-- 
2.31.1

Re: [PATCH] gcc-13: Add changelog for LoongArch.

[Gerald Pfeifer]

First of all thank you! These are quite some nice enhancements, and it's 
great to see them documented.

On Tue, 18 Apr 2023, Lulu Cheng wrote:
> +        <li>The new command-line option <code>-mexplicit-relocs</code> decides whether
> +	    to use or not use the assembler relocation operator when dealing with

Here it is sufficient to say "to use" (and skip "or not to use"), since 
"whether" indicates this kind of choice.

> +	    symbolic addresses. <code>-mexplicit-relocs</code> is enabled by default
> +	    when GCC is configured to use a compatible assembler.

We could say "It" instead of listing the option again. That's just an 
idea, you can keep it as is if you prefer.

> +        </li>
> +        <li>Avoid using the GOT to access external symbols when the new
> +	    <code>-mdirect-extern-access</code> command-line option is specified.
> +        </li>
> +        <li>The <a href="https://gcc.gnu.org/onlinedocs/gcc/LoongArch-Variable-Attributes.html#LoongArch-Variable-Attributes"><code>model</code></a>
> +	    variable attributes has been added.

"attribute" (singular) to match "has".

> +    <li>Built-in functions support Improvements</li>

Maybe just say "Built-in functions"?

> +        <li>The <code>rint</code> and <code>copysign</code> mathematical builtins
> +	    (and their float variants) are now inplemented as inline 

"implemented"

> +	    float variants) are now inplemented as inline LoongArch intrinsics when using

"implemented"

> +	    (and their float variants) are now inplemented as inline LoongArch intrinsics

"implemented"

> +    <li>Subprojects Support</li>
                              ^^^^^
> +      <ul>
> +        <li><code>libvtv</code> now supports LoongArch architecture.</li>
> +        <li><code>libitm</code> now supports LoongArch architecture.</li>
> +	<li>LoongArch supports address sanitizers other than <code>hwasan</code> and <code>tsan</code>.</li>
> +      </ul>

Here, and in the other cases, the closing </li> (that I marked aboved) 
should follow </ul>, since both the heading and the <ul> are part of the
same list item.

(See the RISC-V entry, for example.)


This change is fine with the changes highlighted above. (If you prefer, I 
can have another look; it's not required, though.)

Gerald

Re: [PATCH] gcc-13: Add changelog for LoongArch.

[Lulu Cheng]

在 2023/4/18 下午2:44, Gerald Pfeifer 写道:
> Here, and in the other cases, the closing </li> (that I marked aboved)
> should follow </ul>, since both the heading and the <ul> are part of the
> same list item.
>
> (See the RISC-V entry, for example.)
>
>
> This change is fine with the changes highlighted above. (If you prefer, I
> can have another look; it's not required, though.)
>
> Gerald

I have modified it and published the v2 version, please review it again.

Thanks!

Re: [PATCH] gcc-13: Add changelog for LoongArch.

[WANG Xuerui]

Hi,

Just some minor fixes ;-)

On 2023/4/18 14:15, Lulu Cheng wrote:
> ---
>   htdocs/gcc-13/changes.html | 39 ++++++++++++++++++++++++++++++++++++++
>   1 file changed, 39 insertions(+)
>
> diff --git a/htdocs/gcc-13/changes.html b/htdocs/gcc-13/changes.html
> index f3b9afed..c75e341b 100644
> --- a/htdocs/gcc-13/changes.html
> +++ b/htdocs/gcc-13/changes.html
> @@ -563,6 +563,45 @@ a work-in-progress.</p>
>     </li>
>   </ul>
>   
> +<h3 id="loongarch">LoongArch</h3>
> +  <ul>
> +    <li>New features</li>
> +      <ul>
> +        <li>The new command-line option <code>-mexplicit-relocs</code> decides whether
> +	    to use or not use the assembler relocation operator when dealing with
"provides control over whether ..."?
> +	    symbolic addresses. <code>-mexplicit-relocs</code> is enabled by default
> +	    when GCC is configured to use a compatible assembler.
"if a compatible assembler (binutils 2.40 or later) is present at GCC 
build time"?
> +        </li>
> +        <li>Avoid using the GOT to access external symbols when the new
> +	    <code>-mdirect-extern-access</code> command-line option is specified.

"The new command-line option <code>-mdirect-extern-access</code> can be 
used to prevent accessing external symbols through GOT"?

> +        </li>
> +        <li>The <a href="https://gcc.gnu.org/onlinedocs/gcc/LoongArch-Variable-Attributes.html#LoongArch-Variable-Attributes"><code>model</code></a>
> +	    variable attributes has been added.
"A new variable attribute ... has been added."
> +        </li>
> +      </ul>
> +    <li>Built-in functions support Improvements</li>
"Improvements to built-in functions" or just "Built-in functions" as 
Gerald pointed out.
> +      <ul>
> +        <li>The <code>rint</code> and <code>copysign</code> mathematical builtins
> +	    (and their float variants) are now inplemented as inline LoongArch intrinsics.
> +        </li>
> +        <li>The <code>lrint</code>, <code>logb</code>, <code>scalbln</code>,
> +	    <code>scalbn</code> and <code>ldexp</code> mathematical builtins (and their
> +	    float variants) are now inplemented as inline LoongArch intrinsics when using
> +	    <code>-fno-math-errno</code>.
> +        </li>
> +        <li>The <code>lceil</code> and <code>lfloor</code> mathematical builtins
> +	    (and their float variants) are now inplemented as inline LoongArch intrinsics
> +	    when using <code>-ffp-int-builtin-inexact</code>.
For this part I agree with what Gerald has suggested, the expressions 
are kinda natural after you fix the misspelled "implemented"'s.
> +        </li>
> +      </ul>
> +    <li>Subprojects Support</li>
> +      <ul>
> +        <li><code>libvtv</code> now supports LoongArch architecture.</li>
> +        <li><code>libitm</code> now supports LoongArch architecture.</li>
Remove "architecture"? Because "LoongArch" practically just means 
"Loongson Architecture".
> +	<li>LoongArch supports address sanitizers other than <code>hwasan</code> and <code>tsan</code>.</li>
> +      </ul>
"Address sanitizers other than HWASan and TSan are now supported on 
LoongArch"?
> +  </ul>
> +
>   <!-- <h3 id="mips">MIPS</h3> -->
>   
>   <!-- <h3 id="mep">MeP</h3> -->

Re: [PATCH] gcc-13: Add changelog for LoongArch.

[Lulu Cheng]

在 2023/4/18 下午3:29, WANG Xuerui 写道:
> Hi,
>
> Just some minor fixes ;-)
>
> On 2023/4/18 14:15, Lulu Cheng wrote:
>> ---
>>   htdocs/gcc-13/changes.html | 39 ++++++++++++++++++++++++++++++++++++++
>>   1 file changed, 39 insertions(+)
>>
>> diff --git a/htdocs/gcc-13/changes.html b/htdocs/gcc-13/changes.html
>> index f3b9afed..c75e341b 100644
>> --- a/htdocs/gcc-13/changes.html
>> +++ b/htdocs/gcc-13/changes.html
>> @@ -563,6 +563,45 @@ a work-in-progress.</p>
>>     </li>
>>   </ul>
>>   +<h3 id="loongarch">LoongArch</h3>
>> +  <ul>
>> +    <li>New features</li>
>> +      <ul>
>> +        <li>The new command-line option 
>> <code>-mexplicit-relocs</code> decides whether
>> +        to use or not use the assembler relocation operator when 
>> dealing with
> "provides control over whether ..."?
>> +        symbolic addresses. <code>-mexplicit-relocs</code> is 
>> enabled by default
>> +        when GCC is configured to use a compatible assembler.
> "if a compatible assembler (binutils 2.40 or later) is present at GCC 
> build time"?
>> +        </li>
>> +        <li>Avoid using the GOT to access external symbols when the new
>> +        <code>-mdirect-extern-access</code> command-line option is 
>> specified.
>
> "The new command-line option <code>-mdirect-extern-access</code> can 
> be used to prevent accessing external symbols through GOT"?
>
>> +        </li>
>> +        <li>The <a 
>> href="https://gcc.gnu.org/onlinedocs/gcc/LoongArch-Variable-Attributes.html#LoongArch-Variable-Attributes"><code>model</code></a>
>> +        variable attributes has been added.
> "A new variable attribute ... has been added."
>> +        </li>
>> +      </ul>
>> +    <li>Built-in functions support Improvements</li>
> "Improvements to built-in functions" or just "Built-in functions" as 
> Gerald pointed out.
>> +      <ul>
>> +        <li>The <code>rint</code> and <code>copysign</code> 
>> mathematical builtins
>> +        (and their float variants) are now inplemented as inline 
>> LoongArch intrinsics.
>> +        </li>
>> +        <li>The <code>lrint</code>, <code>logb</code>, 
>> <code>scalbln</code>,
>> +        <code>scalbn</code> and <code>ldexp</code> mathematical 
>> builtins (and their
>> +        float variants) are now inplemented as inline LoongArch 
>> intrinsics when using
>> +        <code>-fno-math-errno</code>.
>> +        </li>
>> +        <li>The <code>lceil</code> and <code>lfloor</code> 
>> mathematical builtins
>> +        (and their float variants) are now inplemented as inline 
>> LoongArch intrinsics
>> +        when using <code>-ffp-int-builtin-inexact</code>.
> For this part I agree with what Gerald has suggested, the expressions 
> are kinda natural after you fix the misspelled "implemented"'s.
>> +        </li>
>> +      </ul>
>> +    <li>Subprojects Support</li>
>> +      <ul>
>> +        <li><code>libvtv</code> now supports LoongArch 
>> architecture.</li>
>> +        <li><code>libitm</code> now supports LoongArch 
>> architecture.</li>
> Remove "architecture"? Because "LoongArch" practically just means 
> "Loongson Architecture".
>> +    <li>LoongArch supports address sanitizers other than 
>> <code>hwasan</code> and <code>tsan</code>.</li>
>> +      </ul>
> "Address sanitizers other than HWASan and TSan are now supported on 
> LoongArch"?
>> +  </ul>
>> +
>>   <!-- <h3 id="mips">MIPS</h3> -->
>>     <!-- <h3 id="mep">MeP</h3> -->

I will continue to improve.


Thanks!:-)