This is the mail archive of the
mailing list for the glibc project.
Re: Remove pre-2.2 Linux kernel support
- From: "Joseph S. Myers" <joseph at codesourcery dot com>
- To: Roland McGrath <roland at hack dot frob dot com>
- Cc: libc-alpha at sourceware dot org, "Ryan S. Arnold" <rsa at us dot ibm dot com>, Thomas Schwinge <thomas at codesourcery dot com>, Kaz Kojima <kkojima at rr dot iij4u dot or dot jp>, "David S. Miller" <davem at davemloft dot net>
- Date: Tue, 8 May 2012 20:06:33 +0000 (UTC)
- Subject: Re: Remove pre-2.2 Linux kernel support
- References: <Pine.LNX.email@example.com><20120508195503.B201D2C0B6@topped-with-meat.com>
On Tue, 8 May 2012, Roland McGrath wrote:
> The thrust of this is certainly fine. I only skimmed the patch.
> One thing I noticed is at least one case (chown) that looks like
> it could be replaced with a syscalls.list entry now.
Yes, as I said there are plenty of further cleanups possible. I suspect
removing the old bounded pointers support (on the list for after 2.16)
could open up more cleanup opportunities as well.
The version of chown that's very simple now is the powerpc one, and I'd
certainly encourage architecture maintainers to do their own cleanups,
looking for conditional code that no longer needs to be conditional or
architecture-specific versions where architecture-independent C files or
syscalls.list entries would now suffice. (Architecture maintainers are
also free to increase the minimum kernel version for their architecture at
their own discretion.) x86 chown will be simplified once we clean up
__ASSUME_32BITUIDS by removing 2.2 and 2.3 kernel support.
Joseph S. Myers