[webkit-reviews] review granted: [Bug 206044] [bmalloc] Calculate LineMetadata for specific VM page size in compile time : [Attachment 387351] PATCH

[bugzilla-daemon at webkit.org]

Yusuke Suzuki <ysuzuki at apple.com> has granted Basuke Suzuki
<Basuke.Suzuki at sony.com>'s request for review:
Bug 206044: [bmalloc] Calculate LineMetadata for specific VM page size in
compile time
https://bugs.webkit.org/show_bug.cgi?id=206044

Attachment 387351: PATCH

https://bugs.webkit.org/attachment.cgi?id=387351&action=review




--- Comment #5 from Yusuke Suzuki <ysuzuki at apple.com> ---
Comment on attachment 387351
  --> https://bugs.webkit.org/attachment.cgi?id=387351
PATCH

View in context: https://bugs.webkit.org/attachment.cgi?id=387351&action=review

r=me so long as we have fallback mechanism which has dynamic calculation b/c
some weird CPU has some weird page size (PowerPC has 64KB).
But I also want to have some comments from Geoff about page size assumption. If
we can dynamically change the page size (maybe we don't especially in macOS /
iOS), we should use dynamic calculation version.

> Source/bmalloc/bmalloc/BPlatform.h:270
> +#if BPLATFORM(iOS) || BPLATFORM(PLAYSTATION)

Typo, BPLATFORM(iOS) => BPLATFORM(IOS). BTW, I'm not sure whether all iOS uses
16KB pages.
@Geoff, do we use 16KB for all iOS (not including watchOS, tvOS and Catalyst)?

For macOS, I think 4KB is OK. Hugepage exists, but from the point of
application, anyway unit of page size is 4KB.