Message 333876 - Python tracker

Message333876

Author	vstinner
Recipients	David.Edelsohn, cstratak, fweimer, mark.dickinson, skrah, vstinner
Date	2019-01-17.14:36:57
SpamBayes Score	-1.0
Marked as misclassified	Yes
Message-id	<1547735817.2.0.908903596667.issue35752@roundup.psfhosted.org>
In-reply-to

Content
> If gcc-8.0.1-0.14.fc28.ppc64le miscompiles memcpy(), perhaps the upstream priority in https://bugzilla.redhat.com/show_bug.cgi?id=1540995 should be "release blocker". With the bug, memoryview doesn't round properly float: rounds to zero rather than rounding to nearest. It's not a critical crash or critical security vulnerability :-) Anyway, I wrote a fix: PR 11593. My fix is to workaround the GCC compiler bug using volatile but only on ppc64. My change only impacts the 'f' type on ppc64 when compiled by GCC. Other architectures, compilers and types are unaffected and so have no impact on performances.

Content

> If gcc-8.0.1-0.14.fc28.ppc64le miscompiles memcpy(), perhaps the upstream priority in https://bugzilla.redhat.com/show_bug.cgi?id=1540995 should be "release blocker".

With the bug, memoryview doesn't round properly float: rounds to zero rather than rounding to nearest. It's not a critical crash or critical security vulnerability :-)

Anyway, I wrote a fix: PR 11593. My fix is to workaround the GCC compiler bug using volatile but only on ppc64. My change only impacts the 'f' type on ppc64 when compiled by GCC. Other architectures, compilers and types are unaffected and so have no impact on performances.

History
Date	User	Action	Args
2019-01-17 14:36:59	vstinner	set	recipients: + vstinner, mark.dickinson, skrah, fweimer, David.Edelsohn, cstratak
2019-01-17 14:36:57	vstinner	set	messageid: <1547735817.2.0.908903596667.issue35752@roundup.psfhosted.org>
2019-01-17 14:36:57	vstinner	link	issue35752 messages
2019-01-17 14:36:57	vstinner	create