Bug 114970 – StringImpl.h should compile with -Wshorten-64-to-32

• Status: RESOLVED
• Resolution: FIXED
• Component: Web Template Framework

[David Kilzer (:ddkilzer)]

Fixes the following warnings with -Wshorten-64-to-32:

    StringImpl.h:317:25: error: implicit conversion loses integer precision: 'uintptr_t' (aka 'unsigned long') to 'unsigned int' [-Werror,-Wshorten-64-to-32]
            unsigned hash = reinterpret_cast<uintptr_t>(this);
                     ~~~~   ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

[David Kilzer (:ddkilzer)]

Created attachment 199053 [details]
Patch v1

[Darin Adler]

Comment on attachment 199053 [details]
Patch v1

View in context: https://bugs.webkit.org/attachment.cgi?id=199053&action=review

> Source/WTF/wtf/text/StringImpl.h:316
> -        unsigned hash = reinterpret_cast<uintptr_t>(this);
> +        unsigned hash = static_cast<uint32_t>(reinterpret_cast<uintptr_t>(this));

This is a silly approach to creating the hash. For example, the low bits will always be due to alignment zero. I suggest we use a random number instead.

Code change is OK, but code is dumb.

[WebKit Commit Bot]

The commit-queue encountered the following flaky tests while processing attachment 199053 [details]:

platform/mac/editing/deleting/deletionUI-single-instance.html bug 114181 (author: [e-mail])
transitions/color-transition-rounding.html bug 114182 (author: [e-mail])
transitions/cubic-bezier-overflow-svg-length.html bug 114183 (author: [e-mail])
transitions/interrupt-zero-duration.html bug 114184 (authors: [e-mail], [e-mail], and [e-mail])
transitions/multiple-background-transitions.html bug 114185 (author: [e-mail])
transitions/cubic-bezier-overflow-color.html bug 114186 (author: [e-mail])
transitions/multiple-shadow-transitions.html bug 114187 (author: [e-mail])
transitions/mismatched-shadow-transitions.html bug 114188 (author: [e-mail])
transitions/color-transition-all.html bug 114189 (authors: [e-mail] and [e-mail])
transitions/negative-delay.html bug 114190 (author: [e-mail])
transitions/cubic-bezier-overflow-shadow.html bug 114191 (author: [e-mail])
transitions/min-max-width-height-transitions.html bug 114192 (author: [e-mail])
transitions/cancel-transition.html bug 114193 (authors: [e-mail], [e-mail], and [e-mail])
transitions/border-radius-transition.html bug 114194 (author: [e-mail])
transitions/flex-transitions.html bug 114195 (author: [e-mail])
transitions/mixed-type.html bug 114196 (author: [e-mail])
transitions/multiple-mask-transitions.html bug 114197 (author: [e-mail])
transitions/color-transition-premultiplied.html bug 114198 (author: [e-mail])
transitions/mismatched-shadow-styles.html bug 114199 (author: [e-mail])
transitions/mask-transitions.html bug 114200 (authors: [e-mail], [e-mail], and [e-mail])
transitions/cubic-bezier-overflow-length.html bug 114201 (author: [e-mail])
transitions/multiple-background-size-transitions.html bug 114202 (authors: [e-mail] and [e-mail])
transitions/clip-transition.html bug 114203 (authors: [e-mail] and [e-mail])
transitions/cubic-bezier-overflow-transform.html bug 114204 (author: [e-mail])
transitions/shorthand-border-transitions.html bug 114205 (authors: [e-mail] and [e-mail])
transitions/interrupted-accelerated-transition.html bug 56242 (authors: [e-mail], [e-mail], and [e-mail])
transitions/background-transitions.html bug 114206 (author: [e-mail])
http/tests/security/cookies/third-party-cookie-blocking-user-action.html bug 114511 (authors: [e-mail], [e-mail], and [e-mail])
http/tests/security/mixedContent/redirect-https-to-http-iframe-in-main-frame.html bug 114208 (authors: [e-mail] and [e-mail])
fast/loader/javascript-url-in-object.html bug 114210 (authors: [e-mail] and [e-mail])
The commit-queue is continuing to process your patch.

[WebKit Commit Bot]

Comment on attachment 199053 [details]
Patch v1

Clearing flags on attachment: 199053

Committed r148900: <http://trac.webkit.org/changeset/148900>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[David Kilzer (:ddkilzer)]

(In reply to comment #2)
> (From update of attachment 199053 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=199053&action=review
> 
> > Source/WTF/wtf/text/StringImpl.h:316
> > -        unsigned hash = reinterpret_cast<uintptr_t>(this);
> > +        unsigned hash = static_cast<uint32_t>(reinterpret_cast<uintptr_t>(this));
> 
> This is a silly approach to creating the hash. For example, the low bits will always be due to alignment zero. I suggest we use a random number instead.
> 
> Code change is OK, but code is dumb.

Should we use WTF::intHash(uint32_t) on 32-bit platforms and WTF::intHas(uint64_t) on 64-bit platforms instead?

[Benjamin Poulain]

(In reply to comment #6)
> (In reply to comment #2)
> > (From update of attachment 199053 [details] [details])
> > View in context: https://bugs.webkit.org/attachment.cgi?id=199053&action=review
> > 
> > > Source/WTF/wtf/text/StringImpl.h:316
> > > -        unsigned hash = reinterpret_cast<uintptr_t>(this);
> > > +        unsigned hash = static_cast<uint32_t>(reinterpret_cast<uintptr_t>(this));
> > 
> > This is a silly approach to creating the hash. For example, the low bits will always be due to alignment zero. I suggest we use a random number instead.
> > 
> > Code change is OK, but code is dumb.
> 
> Should we use WTF::intHash(uint32_t) on 32-bit platforms and WTF::intHas(uint64_t) on 64-bit platforms instead?

The only property needed for that hash is it should never equals the hash of emptyString().

We could basically use a constant here. Any number !m hash(empty()) has the same theoretical chances of conflicts.

I have been trying to kill this constructor every now and then :)

[Darin Adler]

(In reply to comment #7)
> The only property needed for that hash is it should never equals the hash of emptyString().

Maybe (hash(emptyString()) ^ 0x100) would work?

[David Kilzer (:ddkilzer)]

(In reply to comment #8)
> (In reply to comment #7)
> > The only property needed for that hash is it should never equals the hash of emptyString().
> 
> Maybe (hash(emptyString()) ^ 0x100) would work?

Filed Bug 115008 to track this issue.