[webkit-reviews] review granted: [Bug 238726] [Unix] Add UnixFileDescriptor, use it in IPC::Semaphore : [Attachment 456681] Patch
bugzilla-daemon at webkit.org
bugzilla-daemon at webkit.org
Tue Apr 5 02:03:29 PDT 2022
Carlos Garcia Campos <cgarcia at igalia.com> has granted Zan Dobersek
<zan at falconsigh.net>'s request for review:
Bug 238726: [Unix] Add UnixFileDescriptor, use it in IPC::Semaphore
https://bugs.webkit.org/show_bug.cgi?id=238726
Attachment 456681: Patch
https://bugs.webkit.org/attachment.cgi?id=456681&action=review
--- Comment #7 from Carlos Garcia Campos <cgarcia at igalia.com> ---
Comment on attachment 456681
--> https://bugs.webkit.org/attachment.cgi?id=456681
Patch
View in context: https://bugs.webkit.org/attachment.cgi?id=456681&action=review
> Source/WTF/wtf/unix/UnixFileDescriptor.h:51
> + enum DuplicationTag { Duplicate };
> + UnixFileDescriptor(int fd, DuplicationTag)
> + {
> + if (fd >= 0)
> + m_value = dupCloseOnExec(fd);
> + }
I think we can keep this private for now, since it's always used from
::duplicate().
> Source/WTF/wtf/unix/UnixFileDescriptor.h:88
> +
using WTF::UnixFileDescriptor
> Source/WebKit/Platform/IPC/unix/IPCSemaphoreUnix.cpp:44
> + m_fd = { eventfd(0, EFD_CLOEXEC | EFD_NONBLOCK | EFD_SEMAPHORE),
WTF::UnixFileDescriptor::Adopt };
I still prefer m_fd = adoptFileDescriptor(eventfd(0, EFD_CLOEXEC | EFD_NONBLOCK
| EFD_SEMAPHORE)); but no strong opinion if you really want to expose
WTF::UnixFileDescriptor::Adopt and WTF::UnixFileDescriptor::Duplicate
> Source/WebKit/Shared/WebCoreArgumentCoders.h:854
> +#if USE(UNIX_DOMAIN_SOCKETS)
> +
> +template<> struct ArgumentCoder<WTF::UnixFileDescriptor> {
> + static void encode(Encoder&, const WTF::UnixFileDescriptor&);
> + static void encode(Encoder&, WTF::UnixFileDescriptor&&);
> + static std::optional<WTF::UnixFileDescriptor> decode(Decoder&);
> +};
> +
> +#endif
This doesn't really belong here, since UnixFileDescriptor is not a WebCore
class. Why don't you add the encode/decode methods to UnixFileDescriptor class?
More information about the webkit-reviews
mailing list