From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-14.4 required=3.0 tests=DKIMWL_WL_MED,DKIM_SIGNED, DKIM_VALID,DKIM_VALID_AU,HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_PATCH, MAILING_LIST_MULTI,SIGNED_OFF_BY,SPF_HELO_NONE,SPF_PASS,USER_IN_DEF_DKIM_WL autolearn=unavailable autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id BB1DCC3404E for ; Wed, 19 Feb 2020 07:54:49 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 8BE6021D56 for ; Wed, 19 Feb 2020 07:54:49 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="RCM/nYbH" Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1726645AbgBSHys (ORCPT ); Wed, 19 Feb 2020 02:54:48 -0500 Received: from mail-oi1-f193.google.com ([209.85.167.193]:35944 "EHLO mail-oi1-f193.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726582AbgBSHyq (ORCPT ); Wed, 19 Feb 2020 02:54:46 -0500 Received: by mail-oi1-f193.google.com with SMTP id c16so22956613oic.3 for ; Tue, 18 Feb 2020 23:54:46 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20161025; h=mime-version:references:in-reply-to:from:date:message-id:subject:to :cc; bh=U5zuuxTOmeY03tgnTeqmbMJcmo8FWeVE9jhVJeIKxhY=; b=RCM/nYbHYhF26ml+bw7CJB00Tu1bvizUk/NqVAQvp9vRXCFdKYgsl01pnwx5ej8ypI 1YlAzlD1fYBTy6Oc2DhzQUDrhTLVFXLSgs31eQJokjPLkhtO2+8CkEbV5SZjyp5LNEme ziwwWEcQ430fUoAvokqDahZVU5hPf7/uvfR+1bYp/xNUAMXFaDXiD0hkMz15m8P7kWDm SVYKRLaENNJ6LNrWEUk1Zxh3fCOsIYMG1JeA0ttkbAVmmFd3dpbVcxKM39oe1BrLvPhy wqqi+JG0sUJqgbEuJIDFEEd+mU26HfLdO+3TR++rsfNG8L3CtDsdH2AUTmbQAlk4HPnN rE0A== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:mime-version:references:in-reply-to:from:date :message-id:subject:to:cc; bh=U5zuuxTOmeY03tgnTeqmbMJcmo8FWeVE9jhVJeIKxhY=; b=NpX6HES6KDSXDl7hqfL8sNoTIzO+KpSt01Qk+q84dOryDPY0HNLZRL/zenmFNdqySA 0+9OxIJURBTKicb2KQdYsm4Vqb+JefMCr7EqXpItmCHwa1V2TT8thUG/fM67ZZqf7T8V ulfBXc1kEfoy/cb3BG7DzTyFU2vD/xa+2CEWArxzCAgUFjbE26b0MaVxUrkQ0hpl0+fN 8MFN4QiOyPAo+xn2hC34zenwbp0N65gHMq9wlnpX4OyhBfb9tS6WJ3w68k6696i/9io1 GKqtd2+0z5GQASQuyN7S61id2s27NVTckX7xWof05rxhpkzkkJpbEgXqu5gLUq1gcGtH ZkJA== X-Gm-Message-State: APjAAAWgU/mPB6RkN/Wg7IHVIUKTNpJyluh2taPcOv+0gFcmMuNrPkSn qSsCuJG5n2eR01tWAIMW5nGHvkb6MwuuIdPtecrUKw== X-Google-Smtp-Source: APXvYqwrgPiaRBC36d3nGupFESCf7i3fbETUfZ+IXTIAh32oyWQQePytvF+stlAyZxSK3BBnvotq6eC5R+C5XYPzsDs= X-Received: by 2002:aca:d6c8:: with SMTP id n191mr4046734oig.103.1582098885598; Tue, 18 Feb 2020 23:54:45 -0800 (PST) MIME-Version: 1.0 References: <20200211213128.73302-1-almasrymina@google.com> <20200211213128.73302-7-almasrymina@google.com> In-Reply-To: From: Mina Almasry Date: Tue, 18 Feb 2020 23:54:34 -0800 Message-ID: Subject: Re: [PATCH v12 7/9] hugetlb: support file_region coalescing again To: Mike Kravetz Cc: shuah , David Rientjes , Shakeel Butt , Greg Thelen , Andrew Morton , open list , linux-mm@kvack.org, linux-kselftest@vger.kernel.org, cgroups@vger.kernel.org Content-Type: text/plain; charset="UTF-8" Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Tue, Feb 18, 2020 at 7:31 PM Mike Kravetz wrote: > > On 2/11/20 1:31 PM, Mina Almasry wrote: > > An earlier patch in this series disabled file_region coalescing in order > > to hang the hugetlb_cgroup uncharge info on the file_region entries. > > > > This patch re-adds support for coalescing of file_region entries. > > Essentially everytime we add an entry, we call a recursive function that > > tries to coalesce the added region with the regions next to it. The > > worst case call depth for this function is 3: one to coalesce with the > > region next to it, one to coalesce to the region prev, and one to reach > > the base case. > > > > This is an important performance optimization as private mappings add > > their entries page by page, and we could incur big performance costs for > > large mappings with lots of file_region entries in their resv_map. > > > > Signed-off-by: Mina Almasry > > > > --- > > > > Changes in v12: > > - Changed logic for coalescing. Instead of checking inline to coalesce > > with only the region on next or prev, we now have a recursive function > > that takes care of coalescing in both directions. > > - For testing this code I added a bunch of debug code that checks that > > the entries in the resv_map are coalesced appropriately. This passes > > with libhugetlbfs tests. > > > > --- > > mm/hugetlb.c | 85 ++++++++++++++++++++++++++++++++++++++++++++++++++++ > > 1 file changed, 85 insertions(+) > > > > diff --git a/mm/hugetlb.c b/mm/hugetlb.c > > index 2d62dd35399db..45219cb58ac71 100644 > > --- a/mm/hugetlb.c > > +++ b/mm/hugetlb.c > > @@ -276,6 +276,86 @@ static void record_hugetlb_cgroup_uncharge_info(struct hugetlb_cgroup *h_cg, > > #endif > > } > > > > +static bool has_same_uncharge_info(struct file_region *rg, > > + struct file_region *org) > > +{ > > +#ifdef CONFIG_CGROUP_HUGETLB > > + return rg && org && > > + rg->reservation_counter == org->reservation_counter && > > + rg->css == org->css; > > + > > +#else > > + return true; > > +#endif > > +} > > + > > +#ifdef CONFIG_DEBUG_VM > > +static void dump_resv_map(struct resv_map *resv) > > +{ > > + struct list_head *head = &resv->regions; > > + struct file_region *rg = NULL; > > + > > + pr_err("--------- start print resv_map ---------\n"); > > + list_for_each_entry(rg, head, link) { > > + pr_err("rg->from=%ld, rg->to=%ld, rg->reservation_counter=%px, rg->css=%px\n", > > + rg->from, rg->to, rg->reservation_counter, rg->css); > > + } > > + pr_err("--------- end print resv_map ---------\n"); > > +} > > + > > +/* Debug function to loop over the resv_map and make sure that coalescing is > > + * working. > > + */ > > +static void check_coalesce_bug(struct resv_map *resv) > > +{ > > + struct list_head *head = &resv->regions; > > + struct file_region *rg = NULL, *nrg = NULL; > > + > > + list_for_each_entry(rg, head, link) { > > + nrg = list_next_entry(rg, link); > > + > > + if (&nrg->link == head) > > + break; > > + > > + if (nrg->reservation_counter && nrg->from == rg->to && > > + nrg->reservation_counter == rg->reservation_counter && > > + nrg->css == rg->css) { > > + dump_resv_map(resv); > > + VM_BUG_ON(true); > > + } > > + } > > +} > > +#endif > > + > > +static void coalesce_file_region(struct resv_map *resv, struct file_region *rg) > > +{ > > + struct file_region *nrg = NULL, *prg = NULL; > > + > > + prg = list_prev_entry(rg, link); > > + if (&prg->link != &resv->regions && prg->to == rg->from && > > + has_same_uncharge_info(prg, rg)) { > > + prg->to = rg->to; > > + > > + list_del(&rg->link); > > + kfree(rg); > > + > > + coalesce_file_region(resv, prg); > > + return; > > + } > > + > > + nrg = list_next_entry(rg, link); > > + if (&nrg->link != &resv->regions && nrg->from == rg->to && > > + has_same_uncharge_info(nrg, rg)) { > > + nrg->from = rg->from; > > + > > + list_del(&rg->link); > > + kfree(rg); > > + > > + coalesce_file_region(resv, nrg); > > + return; > > + } > > +} > > + > > /* Must be called with resv->lock held. Calling this with count_only == true > > * will count the number of pages to be added but will not modify the linked > > * list. If regions_needed != NULL and count_only == true, then regions_needed > > @@ -327,6 +407,7 @@ static long add_reservation_in_range(struct resv_map *resv, long f, long t, > > record_hugetlb_cgroup_uncharge_info(h_cg, h, > > resv, nrg); > > list_add(&nrg->link, rg->link.prev); > > + coalesce_file_region(resv, nrg); > > } else if (regions_needed) > > *regions_needed += 1; > > } > > @@ -344,11 +425,15 @@ static long add_reservation_in_range(struct resv_map *resv, long f, long t, > > resv, last_accounted_offset, t); > > record_hugetlb_cgroup_uncharge_info(h_cg, h, resv, nrg); > > list_add(&nrg->link, rg->link.prev); > > + coalesce_file_region(resv, nrg); > > } else if (regions_needed) > > *regions_needed += 1; > > } > > > > VM_BUG_ON(add < 0); > > +#ifdef CONFIG_DEBUG_VM > > + check_coalesce_bug(resv); > > +#endif > > Some distros have CONFIG_DEBUG_VM on in their default kernels. Fedora comes > to mind. Yes, this means 'VM_BUG_ON()' become 'BUG_ON()'. That is somewhat > accepted. I don't think we want this debug code behind CONFIG_DEBUG_VM and > called each time a file region is added. It may be overkill to make it a > debug option via the config system. Perhaps, just make it something like > CONFIG_DEBUG_HUGETLB_RSV_REGION and let hugetlb developers turn it on if > they would like? > Ah, I feel like adding a whole config to accommodate this function is overkill and will end up being dead code anyway as no-one remembers to turn it on. If it's the same to you I'll take it out of the patch and leave it as a local change for me to test with. > Other than that, the code looks good. > -- > Mike Kravetz > > > return add; > > } > > > > -- > > 2.25.0.225.g125e21ebc7-goog > >