Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754591AbaGNLdx (ORCPT ); Mon, 14 Jul 2014 07:33:53 -0400 Received: from ozlabs.org ([103.22.144.67]:33338 "EHLO ozlabs.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753761AbaGNLdq (ORCPT ); Mon, 14 Jul 2014 07:33:46 -0400 From: Rusty Russell To: Lucas Tanure , Dave Hansen , linux-kernel@vger.kernel.org Subject: Re: [PATCH] NoMoreModuleVmalloc - Try alloc_pages_exact before vmalloc, if fails do vmalloc In-Reply-To: <20140713130407.GA27344@archDesk> References: <20140713130407.GA27344@archDesk> User-Agent: Notmuch/0.17 (http://notmuchmail.org) Emacs/24.3.1 (x86_64-pc-linux-gnu) Date: Mon, 14 Jul 2014 19:48:46 +0930 Message-ID: <87lhrw4515.fsf@rustcorp.com.au> MIME-Version: 1.0 Content-Type: text/plain Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Lucas Tanure writes: > Convert vmalloc() call to alloc_pages_exact(). If alloc_pages_exact() fails, > then fall back to vmalloc(). If the address is a vmalloc address, then > call vfree(), otherwise call free_pages_exact(). Hi Lucas, This patch is wrong. Firstly, you didn't list a motivation for getting rid of vmalloc. But this patch doesn't make modules avoid vmalloc, it just makes the temporary copy we create avoid vmalloc. It also doesn't avoid the vfree() on the normal path; I'm pretty sure this would crash if you tested it. Thanks, Rusty. > Signed-off-by: Lucas Tanure > --- > kernel/module.c | 16 ++++++++++++---- > 1 file changed, 12 insertions(+), 4 deletions(-) > > diff --git a/kernel/module.c b/kernel/module.c > index ae79ce6..50c1e77 100644 > --- a/kernel/module.c > +++ b/kernel/module.c > @@ -2484,6 +2484,7 @@ static int copy_module_from_user(const void __user *umod, unsigned long len, > struct load_info *info) > { > int err; > + bool alloc_from_vmalloc = false; > > info->len = len; > if (info->len < sizeof(*(info->hdr))) > @@ -2494,12 +2495,19 @@ static int copy_module_from_user(const void __user *umod, unsigned long len, > return err; > > /* Suck in entire file: we'll want most of it. */ > - info->hdr = vmalloc(info->len); > - if (!info->hdr) > - return -ENOMEM; > + info->hdr = alloc_pages_exact(info->len, GFP_KERNEL); > + if (!info->hdr) { > + info->hdr = vmalloc(info->len); > + if (!info->hdr) > + return -ENOMEM; > + alloc_from_vmalloc = true; > + } > > if (copy_from_user(info->hdr, umod, info->len) != 0) { > - vfree(info->hdr); > + if(alloc_from_vmalloc) > + vfree(info->hdr); > + else > + free_pages_exact(info->hdr,info->len); > return -EFAULT; > } > > -- > 2.0.1 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/