From mboxrd@z Thu Jan 1 00:00:00 1970 Content-Type: text/plain; charset="iso-8859-1" From: Daniel Phillips Subject: Re: [RFC] Page table sharing Date: Tue, 19 Feb 2002 03:22:21 +0100 References: In-Reply-To: MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-Id: Sender: owner-linux-mm@kvack.org Return-Path: To: Linus Torvalds , Rik van Riel Cc: Hugh Dickins , dmccr@us.ibm.com, Kernel Mailing List , linux-mm@kvack.org, Robert Love , mingo@redhat.co, Andrew Morton , manfred@colorfullife.com, wli@holomorphy.com List-ID: On February 19, 2002 03:05 am, Linus Torvalds wrote: > On Mon, 18 Feb 2002, Rik van Riel wrote: > > > > The swapout code can remove a page from the page table > > while another process is in the process of unsharing > > the page table. > > Ok, I'll buy that. However, looking at that, the locking is not the real > issue at all: > > When the swapper does a "ptep_get_and_clear()" on a shared pmd, it will > end up having to not just synchronize with anybody doing unsharing, it > will have to flush all the TLB's on all the mm's that might be implicated. > > Which implies that the swapper needs to look up all mm's some way anyway, Ick. With rmap this is straightforward, but without, what? flush_tlb_all? Maybe page tables should be unshared on swapin/out after all, only on arches that need special tlb treatment, or until we have rmap. > so the locking gets solved that way. -- Daniel -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/