From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752065AbdGaMlI (ORCPT ); Mon, 31 Jul 2017 08:41:08 -0400 Received: from mx0b-001b2d01.pphosted.com ([148.163.158.5]:37204 "EHLO mx0a-001b2d01.pphosted.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1751199AbdGaMlH (ORCPT ); Mon, 31 Jul 2017 08:41:07 -0400 Date: Mon, 31 Jul 2017 14:40:53 +0200 From: Gerald Schaefer To: Michal Hocko Cc: linux-mm@kvack.org, Andrew Morton , Mel Gorman , Vlastimil Babka , Andrea Arcangeli , Jerome Glisse , Reza Arbab , Yasuaki Ishimatsu , qiuxishi@huawei.com, Kani Toshimitsu , slaoub@gmail.com, Joonsoo Kim , Andi Kleen , Daniel Kiper , Igor Mammedov , Vitaly Kuznetsov , LKML , Michal Hocko , Benjamin Herrenschmidt , Catalin Marinas , Fenghua Yu , Heiko Carstens , "H. Peter Anvin" , Ingo Molnar , Martin Schwidefsky , Michael Ellerman , Paul Mackerras , Thomas Gleixner , Tony Luck , Will Deacon , gerald.schaefer@de.ibm.com Subject: Re: [RFC PATCH 2/5] mm, arch: unify vmemmap_populate altmap handling In-Reply-To: <20170726083333.17754-3-mhocko@kernel.org> References: <20170726083333.17754-1-mhocko@kernel.org> <20170726083333.17754-3-mhocko@kernel.org> Organization: IBM Deutschland Research & Development GmbH / Vorsitzende des Aufsichtsrats: Martina Koederitz / Geschaeftsfuehrung: Dirk Wittkopp / Sitz der Gesellschaft: Boeblingen / Registergericht: Amtsgericht Stuttgart, HRB 243294 X-Mailer: Claws Mail 3.9.0 (GTK+ 2.24.23; x86_64-redhat-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-TM-AS-GCONF: 00 x-cbid: 17073112-0040-0000-0000-000003E9F9EE X-IBM-AV-DETECTION: SAVI=unused REMOTE=unused XFE=unused x-cbparentid: 17073112-0041-0000-0000-00002087D716 Message-Id: <20170731144053.38c8b012@thinkpad> X-Proofpoint-Virus-Version: vendor=fsecure engine=2.50.10432:,, definitions=2017-07-31_06:,, signatures=0 X-Proofpoint-Spam-Details: rule=outbound_notspam policy=outbound score=0 spamscore=0 suspectscore=2 malwarescore=0 phishscore=0 adultscore=0 bulkscore=0 classifier=spam adjust=0 reason=mlx scancount=1 engine=8.0.1-1706020000 definitions=main-1707310215 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed, 26 Jul 2017 10:33:30 +0200 Michal Hocko wrote: > From: Michal Hocko > > vmem_altmap allows vmemmap_populate to allocate memmap (struct page > array) from an alternative allocator rather than bootmem resp. > kmalloc. Only x86 currently supports altmap handling, most likely > because only nvdim code uses this mechanism currently and the code > depends on ZONE_DEVICE which is present only for x86_64. This will > change in follow up changes so we would like other architectures > to support it as well. > > Provide vmemmap_populate generic implementation which simply resolves > altmap and then call into arch specific __vmemmap_populate. > Architectures then only need to use __vmemmap_alloc_block_buf to > allocate the memmap. vmemmap_free then needs to call vmem_altmap_free > if there is any altmap associated with the address. > > This patch shouldn't introduce any functional changes because > to_vmem_altmap always returns NULL on !x86_x64. > > Cc: Catalin Marinas > Cc: Will Deacon > Cc: Tony Luck > Cc: Fenghua Yu > Cc: Benjamin Herrenschmidt > Cc: Paul Mackerras > Cc: Michael Ellerman > Cc: Martin Schwidefsky > Cc: Heiko Carstens > Cc: Thomas Gleixner > Cc: Ingo Molnar > Cc: "H. Peter Anvin" > Cc: linuxppc-dev@lists.ozlabs.org > Cc: linux-ia64@vger.kernel.org > Cc: x86@kernel.org > Signed-off-by: Michal Hocko > --- > arch/arm64/mm/mmu.c | 9 ++++++--- > arch/ia64/mm/discontig.c | 4 +++- > arch/powerpc/mm/init_64.c | 29 ++++++++++++++++++++--------- > arch/s390/mm/vmem.c | 7 ++++--- > arch/sparc/mm/init_64.c | 6 +++--- > arch/x86/mm/init_64.c | 4 ++-- > include/linux/memremap.h | 13 ++----------- > include/linux/mm.h | 19 ++++++++++++++++++- > mm/sparse-vmemmap.c | 2 +- > 9 files changed, 59 insertions(+), 34 deletions(-) > > diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c > index 0c429ec6fde8..5de1161e7a1b 100644 > --- a/arch/arm64/mm/mmu.c > +++ b/arch/arm64/mm/mmu.c > @@ -649,12 +649,15 @@ int kern_addr_valid(unsigned long addr) > } > #ifdef CONFIG_SPARSEMEM_VMEMMAP > #if !ARM64_SWAPPER_USES_SECTION_MAPS > -int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node) > +int __meminit __vmemmap_populate(unsigned long start, unsigned long end, int node, > + struct vmem_altmap *altmap) > { > + WARN(altmap, "altmap unsupported\n"); > return vmemmap_populate_basepages(start, end, node); > } > #else /* !ARM64_SWAPPER_USES_SECTION_MAPS */ > -int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node) > +int __meminit __vmemmap_populate(unsigned long start, unsigned long end, int node, > + struct vmem_altmap *altmap) > { > unsigned long addr = start; > unsigned long next; > @@ -677,7 +680,7 @@ int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node) > if (pmd_none(*pmd)) { > void *p = NULL; > > - p = vmemmap_alloc_block_buf(PMD_SIZE, node); > + p = __vmemmap_alloc_block_buf(PMD_SIZE, node, altmap); > if (!p) > return -ENOMEM; > > diff --git a/arch/ia64/mm/discontig.c b/arch/ia64/mm/discontig.c > index 878626805369..2a939e877ced 100644 > --- a/arch/ia64/mm/discontig.c > +++ b/arch/ia64/mm/discontig.c > @@ -753,8 +753,10 @@ void arch_refresh_nodedata(int update_node, pg_data_t *update_pgdat) > #endif > > #ifdef CONFIG_SPARSEMEM_VMEMMAP > -int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node) > +int __meminit __vmemmap_populate(unsigned long start, unsigned long end, int node, > + struct vmem_altmap *altmap) > { > + WARN(altmap, "altmap unsupported\n"); > return vmemmap_populate_basepages(start, end, node); > } > > diff --git a/arch/powerpc/mm/init_64.c b/arch/powerpc/mm/init_64.c > index ec84b31c6c86..5ea5e870a589 100644 > --- a/arch/powerpc/mm/init_64.c > +++ b/arch/powerpc/mm/init_64.c > @@ -44,6 +44,7 @@ > #include > #include > #include > +#include > > #include > #include > @@ -115,7 +116,8 @@ static struct vmemmap_backing *next; > static int num_left; > static int num_freed; > > -static __meminit struct vmemmap_backing * vmemmap_list_alloc(int node) > +static __meminit struct vmemmap_backing * vmemmap_list_alloc(int node, > + struct vmem_altmap *altmap) > { > struct vmemmap_backing *vmem_back; > /* get from freed entries first */ > @@ -129,7 +131,7 @@ static __meminit struct vmemmap_backing * vmemmap_list_alloc(int node) > > /* allocate a page when required and hand out chunks */ > if (!num_left) { > - next = vmemmap_alloc_block(PAGE_SIZE, node); > + next = __vmemmap_alloc_block_buf(PAGE_SIZE, node, altmap); > if (unlikely(!next)) { > WARN_ON(1); > return NULL; > @@ -144,11 +146,12 @@ static __meminit struct vmemmap_backing * vmemmap_list_alloc(int node) > > static __meminit void vmemmap_list_populate(unsigned long phys, > unsigned long start, > - int node) > + int node, > + struct vmem_altmap *altmap) > { > struct vmemmap_backing *vmem_back; > > - vmem_back = vmemmap_list_alloc(node); > + vmem_back = vmemmap_list_alloc(node, altmap); > if (unlikely(!vmem_back)) { > WARN_ON(1); > return; > @@ -161,14 +164,15 @@ static __meminit void vmemmap_list_populate(unsigned long phys, > vmemmap_list = vmem_back; > } > > -int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node) > +int __meminit __vmemmap_populate(unsigned long start, unsigned long end, int node, > + struct vmem_altmap *altmap) > { > unsigned long page_size = 1 << mmu_psize_defs[mmu_vmemmap_psize].shift; > > /* Align to the page size of the linear mapping. */ > start = _ALIGN_DOWN(start, page_size); > > - pr_debug("vmemmap_populate %lx..%lx, node %d\n", start, end, node); > + pr_debug("__vmemmap_populate %lx..%lx, node %d\n", start, end, node); > > for (; start < end; start += page_size) { > void *p; > @@ -177,11 +181,11 @@ int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node) > if (vmemmap_populated(start, page_size)) > continue; > > - p = vmemmap_alloc_block(page_size, node); > + p = __vmemmap_alloc_block_buf(page_size, node, altmap); > if (!p) > return -ENOMEM; > > - vmemmap_list_populate(__pa(p), start, node); > + vmemmap_list_populate(__pa(p), start, node, altmap); > > pr_debug(" * %016lx..%016lx allocated at %p\n", > start, start + page_size, p); > @@ -189,7 +193,7 @@ int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node) > rc = vmemmap_create_mapping(start, page_size, __pa(p)); > if (rc < 0) { > pr_warning( > - "vmemmap_populate: Unable to create vmemmap mapping: %d\n", > + "__vmemmap_populate: Unable to create vmemmap mapping: %d\n", > rc); > return -EFAULT; > } > @@ -253,6 +257,12 @@ void __ref vmemmap_free(unsigned long start, unsigned long end) > addr = vmemmap_list_free(start); > if (addr) { > struct page *page = pfn_to_page(addr >> PAGE_SHIFT); > + struct vmem_altmap *altmap = to_vmem_altmap((unsigned long) page); > + > + if (altmap) { > + vmem_altmap_free(altmap, page_size >> PAGE_SHIFT); > + goto unmap; > + } > > if (PageReserved(page)) { > /* allocated from bootmem */ > @@ -272,6 +282,7 @@ void __ref vmemmap_free(unsigned long start, unsigned long end) > free_pages((unsigned long)(__va(addr)), > get_order(page_size)); > > +unmap: > vmemmap_remove_mapping(start, page_size); > } > } > diff --git a/arch/s390/mm/vmem.c b/arch/s390/mm/vmem.c > index c33c94b4be60..07120bc137a1 100644 > --- a/arch/s390/mm/vmem.c > +++ b/arch/s390/mm/vmem.c > @@ -208,7 +208,8 @@ static void vmem_remove_range(unsigned long start, unsigned long size) > /* > * Add a backed mem_map array to the virtual mem_map array. > */ > -int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node) > +int __meminit __vmemmap_populate(unsigned long start, unsigned long end, int node, > + struct vmem_altmap *altmap) > { > unsigned long pgt_prot, sgt_prot; > unsigned long address = start; > @@ -247,12 +248,12 @@ int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node) > * use large frames even if they are only partially > * used. > * Otherwise we would have also page tables since > - * vmemmap_populate gets called for each section > + * __vmemmap_populate gets called for each section > * separately. */ > if (MACHINE_HAS_EDAT1) { > void *new_page; > > - new_page = vmemmap_alloc_block(PMD_SIZE, node); > + new_page = __vmemmap_alloc_block_buf(PMD_SIZE, node, altmap); > if (!new_page) > goto out; > pmd_val(*pm_dir) = __pa(new_page) | sgt_prot; There is another call to vmemmap_alloc_block() in this function, a couple of lines below, this should also be replaced by __vmemmap_alloc_block_buf(). We won't be able to use this directly on s390 yet, but your work should get us a big step closer, thanks. Regards, Gerald