From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-pf0-f199.google.com (mail-pf0-f199.google.com [209.85.192.199]) by kanga.kvack.org (Postfix) with ESMTP id 20C706B0038 for ; Mon, 15 Jan 2018 00:57:06 -0500 (EST) Received: by mail-pf0-f199.google.com with SMTP id n6so9328020pfg.19 for ; Sun, 14 Jan 2018 21:57:06 -0800 (PST) Received: from mail-sor-f65.google.com (mail-sor-f65.google.com. [209.85.220.65]) by mx.google.com with SMTPS id q9sor1365182pfl.92.2018.01.14.21.57.04 for (Google Transport Security); Sun, 14 Jan 2018 21:57:04 -0800 (PST) Date: Sun, 14 Jan 2018 21:57:01 -0800 From: Omar Sandoval Subject: Re: [PATCH 4.14 023/159] mm/sparsemem: Allocate mem_section at runtime for CONFIG_SPARSEMEM_EXTREME=y Message-ID: <20180115055701.GA9071@vader> References: <20180108160444.2ol4fvgqbxnjmlpg@gmail.com> <20180108174653.7muglyihpngxp5tl@black.fi.intel.com> <20180109001303.dy73bpixsaegn4ol@node.shutemov.name> <20180109010927.GA2082@dhcp-128-65.nay.redhat.com> <20180109054131.GB1935@localhost.localdomain> <20180109072440.GA6521@dhcp-128-65.nay.redhat.com> <20180109090552.45ddfk2y25lf4uyn@node.shutemov.name> <20180110030804.GB1744@dhcp-128-110.nay.redhat.com> <20180110111603.56disgew7ipusgjy@black.fi.intel.com> <20180112005549.GA2265@dhcp-128-65.nay.redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20180112005549.GA2265@dhcp-128-65.nay.redhat.com> Sender: owner-linux-mm@kvack.org List-ID: To: Dave Young Cc: "Kirill A. Shutemov" , Andrew Morton , "Kirill A. Shutemov" , Baoquan He , Ingo Molnar , Mike Galbraith , Greg Kroah-Hartman , linux-kernel@vger.kernel.org, stable@vger.kernel.org, Andy Lutomirski , Borislav Petkov , Cyrill Gorcunov , Linus Torvalds , Peter Zijlstra , Thomas Gleixner , linux-mm@kvack.org, Vivek Goyal , kexec@lists.infradead.org On Fri, Jan 12, 2018 at 08:55:49AM +0800, Dave Young wrote: > On 01/10/18 at 02:16pm, Kirill A. Shutemov wrote: > > On Wed, Jan 10, 2018 at 03:08:04AM +0000, Dave Young wrote: > > > On Tue, Jan 09, 2018 at 12:05:52PM +0300, Kirill A. Shutemov wrote: > > > > On Tue, Jan 09, 2018 at 03:24:40PM +0800, Dave Young wrote: > > > > > On 01/09/18 at 01:41pm, Baoquan He wrote: > > > > > > On 01/09/18 at 09:09am, Dave Young wrote: > > > > > > > > > > > > > As for the macro name, VMCOREINFO_SYMBOL_ARRAY sounds better. > > > > > > > > Yep, that's better. > > > > > > > > > > I still think using vmcoreinfo_append_str is better. Unless we replace > > > > > > all array variables with the newly added macro. > > > > > > > > > > > > vmcoreinfo_append_str("SYMBOL(mem_section)=%lx\n", > > > > > > (unsigned long)mem_section); > > > > > > > > > > I have no strong opinion, either change all array uses or just introduce > > > > > the macro and start to use it from now on if we have similar array > > > > > symbols. > > > > > > > > Do you need some action on my side or will you folks take care about this? > > > > > > I think Baoquan was suggesting to update all array users in current > > > code, if you can check every VMCOREINFO_SYMBOL and update all the arrays > > > he will be happy. But if can not do it easily I'm fine with a > > > VMCOREINFO_SYMBOL_ARRAY changes only now, we kdump people can do it > > > later as well. > > > > It seems it's the only array we have there. swapper_pg_dir is a potential > > candidate, but it's 'unsigned long' on arm. > > > > Below it patch with corrected macro name. > > > > Please, consider applying. > > > > From 70f3a84b97f2de98d1364f7b10b7a42a1d8e9968 Mon Sep 17 00:00:00 2001 > > From: "Kirill A. Shutemov" > > Date: Tue, 9 Jan 2018 02:55:47 +0300 > > Subject: [PATCH] kdump: Write a correct address of mem_section into vmcoreinfo > > > > Depending on configuration mem_section can now be an array or a pointer > > to an array allocated dynamically. In most cases, we can continue to refer > > to it as 'mem_section' regardless of what it is. > > > > But there's one exception: '&mem_section' means "address of the array" if > > mem_section is an array, but if mem_section is a pointer, it would mean > > "address of the pointer". > > > > We've stepped onto this in kdump code. VMCOREINFO_SYMBOL(mem_section) > > writes down address of pointer into vmcoreinfo, not array as we wanted. > > > > Let's introduce VMCOREINFO_SYMBOL_ARRAY() that would handle the > > situation correctly for both cases. > > > > Signed-off-by: Kirill A. Shutemov > > Fixes: 83e3c48729d9 ("mm/sparsemem: Allocate mem_section at runtime for CONFIG_SPARSEMEM_EXTREME=y") > > --- > > include/linux/crash_core.h | 2 ++ > > kernel/crash_core.c | 2 +- > > 2 files changed, 3 insertions(+), 1 deletion(-) > > > > diff --git a/include/linux/crash_core.h b/include/linux/crash_core.h > > index 06097ef30449..b511f6d24b42 100644 > > --- a/include/linux/crash_core.h > > +++ b/include/linux/crash_core.h > > @@ -42,6 +42,8 @@ phys_addr_t paddr_vmcoreinfo_note(void); > > vmcoreinfo_append_str("PAGESIZE=%ld\n", value) > > #define VMCOREINFO_SYMBOL(name) \ > > vmcoreinfo_append_str("SYMBOL(%s)=%lx\n", #name, (unsigned long)&name) > > +#define VMCOREINFO_SYMBOL_ARRAY(name) \ > > + vmcoreinfo_append_str("SYMBOL(%s)=%lx\n", #name, (unsigned long)name) > > #define VMCOREINFO_SIZE(name) \ > > vmcoreinfo_append_str("SIZE(%s)=%lu\n", #name, \ > > (unsigned long)sizeof(name)) > > diff --git a/kernel/crash_core.c b/kernel/crash_core.c > > index b3663896278e..4f63597c824d 100644 > > --- a/kernel/crash_core.c > > +++ b/kernel/crash_core.c > > @@ -410,7 +410,7 @@ static int __init crash_save_vmcoreinfo_init(void) > > VMCOREINFO_SYMBOL(contig_page_data); > > #endif > > #ifdef CONFIG_SPARSEMEM > > - VMCOREINFO_SYMBOL(mem_section); > > + VMCOREINFO_SYMBOL_ARRAY(mem_section); > > VMCOREINFO_LENGTH(mem_section, NR_SECTION_ROOTS); > > VMCOREINFO_STRUCT_SIZE(mem_section); > > VMCOREINFO_OFFSET(mem_section, section_mem_map); > > -- > > Kirill A. Shutemov > > > Acked-by: Dave Young > > If stable kernel took the mem section commits, then should also cc > stable. Andrew, can you help to make this in 4.15? > > Thanks > Dave Hm, this fix means that the vmlinux symbol table and vmcoreinfo have different values for mem_section. That seems... odd. I had to patch makedumpfile to fix the case of an explicit vmlinux being passed on the command line (which I realized I don't need to do, but it should still work):