From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from sc8-sf-mx1-b.sourceforge.net ([10.3.1.91] helo=mail.sourceforge.net) by sc8-sf-list1.sourceforge.net with esmtp (Exim 4.30) id 1EBsni-000763-UZ for user-mode-linux-devel@lists.sourceforge.net; Sun, 04 Sep 2005 04:36:42 -0700 Received: from smtp001.mail.ukl.yahoo.com ([217.12.11.32]) by mail.sourceforge.net with smtp (Exim 4.44) id 1EBsnf-0006Nh-Ep for user-mode-linux-devel@lists.sourceforge.net; Sun, 04 Sep 2005 04:36:43 -0700 From: Blaisorblade Subject: Re: [uml-devel] Re: [patch 1/3] uml: share page bits handling between 2 and 3 level pagetables References: <20050728185655.9C6ADA3@zion.home.lan> <200508102137.28414.blaisorblade@yahoo.it> <20050902201749.GA9104@ccure.user-mode-linux.org> In-Reply-To: <20050902201749.GA9104@ccure.user-mode-linux.org> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200509041333.30466.blaisorblade@yahoo.it> Sender: user-mode-linux-devel-admin@lists.sourceforge.net Errors-To: user-mode-linux-devel-admin@lists.sourceforge.net List-Unsubscribe: , List-Id: The user-mode Linux development list List-Post: List-Help: List-Subscribe: , List-Archive: Date: Sun, 4 Sep 2005 13:33:30 +0200 To: user-mode-linux-devel@lists.sourceforge.net Cc: Jeff Dike , linux-kernel@vger.kernel.org On Friday 02 September 2005 22:17, Jeff Dike wrote: > On Wed, Aug 10, 2005 at 09:37:28PM +0200, Blaisorblade wrote: > > Also look, on the "set_pte" theme, at the attached patch. > + WARN_ON(!pte_young(*pte) || pte_write(*pte) && !pte_dirty(*pte)); > This one has been firing on me, and I decided to figure out why. The > culprit is this code in do_no_page: > if (pte_none(*page_table)) { > if (!PageReserved(new_page)) > inc_mm_counter(mm, rss); > > flush_icache_page(vma, new_page); > entry = mk_pte(new_page, vma->vm_page_prot); > if (write_access) > entry = maybe_mkwrite(pte_mkdirty(entry), vma); > set_pte_at(mm, address, page_table, entry); > > The first mk_pte immediately sets the pte to the protection limits of > the VMA, regardless of the access type. > So, if it's a read access on > a writeable page, we get a writeable, but not dirty pte, since the > mkdirty never happens. The exercises the warning you added. Thanks for noticing - I had really this doubt when writing some code (in the patch, I've added a dirty PTEs on read accesses because I was unsure, and even because of my warning). > This seems somewhat bogus to me. If we set the pte protection to its > limits, then the maybe_mkwrite is unneccesary. > This doesn't seem to harm our dirty bit emulation. fix_range_common > checks the dirty and accessed bits and disables read and write > protection as appropriate. > So, it seems like the warning could be dropped, or perhaps made more > selective, like checking for is_write == 0 and VM_WRITE, but then the > test is getting complicated. No, just replace pte_write() with is_write, as below. They might not coincide, but if on a write fault we return with a clean PTE, we'll loop indefinitely (experienced while hacking on remap_f_p), so the warning above is definitely correct. WARN_ON(!pte_young(*pte) || is_write && !pte_dirty(*pte)); > Heff -- Inform me of my mistakes, so I can keep imitating Homer Simpson's "Doh!". Paolo Giarrusso, aka Blaisorblade (Skype ID "PaoloGiarrusso", ICQ 215621894) http://www.user-mode-linux.org/~blaisorblade ___________________________________ Yahoo! Messenger: chiamate gratuite in tutto il mondo http://it.beta.messenger.yahoo.com ------------------------------------------------------- SF.Net email is Sponsored by the Better Software Conference & EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf _______________________________________________ User-mode-linux-devel mailing list User-mode-linux-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel