From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-pf0-x241.google.com (mail-pf0-x241.google.com [IPv6:2607:f8b0:400e:c00::241]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by lists.ozlabs.org (Postfix) with ESMTPS id 3yVrKJ3XBYzDrKq for ; Mon, 6 Nov 2017 22:42:59 +1100 (AEDT) Received: by mail-pf0-x241.google.com with SMTP id a8so7553776pfc.0 for ; Mon, 06 Nov 2017 03:42:59 -0800 (PST) Date: Mon, 6 Nov 2017 22:42:46 +1100 From: Nicholas Piggin To: "Aneesh Kumar K.V" Cc: linuxppc-dev@lists.ozlabs.org, Florian Weimer Subject: Re: [PATCH 4/5] powerpc/64s/radix: Fix 128TB-512TB virtual address boundary case allocation Message-ID: <20171106224246.05f234c5@roar.ozlabs.ibm.com> In-Reply-To: <87po8vslp3.fsf@linux.vnet.ibm.com> References: <20171106100315.29720-1-npiggin@gmail.com> <20171106100315.29720-5-npiggin@gmail.com> <87po8vslp3.fsf@linux.vnet.ibm.com> MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Mon, 06 Nov 2017 16:44:48 +0530 "Aneesh Kumar K.V" wrote: > Nicholas Piggin writes: > > > Radix VA space allocations test addresses against mm->task_size which is > > 512TB, even in cases where the intention is to limit allocation to below > > 128TB. > > > > This results in mmap with a hint address below 128TB but address + length > > above 128TB succeeding when it should fail (as hash does after the > > previous patch). > > > > Set the high address limit to be considered up front, and base subsequent > > allocation checks on that consistently. > > Doesn't setting info.high_limit take care of that ? I would expect > vm_unmapped_area to fail based on info.high_limit. No, it is the hint address case. info.high_limit only gets involved if the hint area was unavailable. I prefer the behaviour without this fix because I disagree that the explicit address request should fail, but this is what you asked for. Actually now I come to look again, it seems that generic code does *not* fail in this case either! Any explicit hint will succeed if it partially or completely crosses 128TB. This is much better behaviour, so I think powerpc has it wrong. > Is this with MAP_FIXED? With MAP_FIXED, it remains as succeeding as expected (like generic code and hash). I did not change that case. > > > > > > Cc: "Aneesh Kumar K.V" > > Fixes: f4ea6dcb08 ("powerpc/mm: Enable mappings above 128TB") > > Signed-off-by: Nicholas Piggin > > --- > > arch/powerpc/mm/hugetlbpage-radix.c | 13 +++++++------ > > arch/powerpc/mm/mmap.c | 27 ++++++++++++++------------- > > 2 files changed, 21 insertions(+), 19 deletions(-) > > > > diff --git a/arch/powerpc/mm/hugetlbpage-radix.c b/arch/powerpc/mm/hugetlbpage-radix.c > > index a12e86395025..9c6a411e9c85 100644 > > --- a/arch/powerpc/mm/hugetlbpage-radix.c > > +++ b/arch/powerpc/mm/hugetlbpage-radix.c > > @@ -48,14 +48,18 @@ radix__hugetlb_get_unmapped_area(struct file *file, unsigned long addr, > > struct mm_struct *mm = current->mm; > > struct vm_area_struct *vma; > > struct hstate *h = hstate_file(file); > > + unsigned long high_limit = DEFAULT_MAP_WINDOW; > > struct vm_unmapped_area_info info; > > > > if (unlikely(addr > mm->context.addr_limit && addr < TASK_SIZE)) > > mm->context.addr_limit = TASK_SIZE; > > > > + if (addr > high_limit) > > + high_limit = TASK_SIZE; > > + > > if (len & ~huge_page_mask(h)) > > return -EINVAL; > > - if (len > mm->task_size) > > + if (len > high_limit) > > return -ENOMEM; > > > > if (flags & MAP_FIXED) { > > @@ -67,7 +71,7 @@ radix__hugetlb_get_unmapped_area(struct file *file, unsigned long addr, > > if (addr) { > > addr = ALIGN(addr, huge_page_size(h)); > > vma = find_vma(mm, addr); > > - if (mm->task_size - len >= addr && > > + if (high_limit - len >= addr && > > (!vma || addr + len <= vm_start_gap(vma))) > > return addr; > > } > > @@ -78,12 +82,9 @@ radix__hugetlb_get_unmapped_area(struct file *file, unsigned long addr, > > info.flags = VM_UNMAPPED_AREA_TOPDOWN; > > info.length = len; > > info.low_limit = PAGE_SIZE; > > - info.high_limit = current->mm->mmap_base; > > + info.high_limit = mm->mmap_base + (high_limit - DEFAULT_MAP_WINDOW); > > info.align_mask = PAGE_MASK & ~huge_page_mask(h); > > info.align_offset = 0; > > > > - if (addr > DEFAULT_MAP_WINDOW) > > - info.high_limit += mm->context.addr_limit - DEFAULT_MAP_WINDOW; > > - > > return vm_unmapped_area(&info); > > } > > diff --git a/arch/powerpc/mm/mmap.c b/arch/powerpc/mm/mmap.c > > index 5d78b193fec4..e6cb3b3f7e93 100644 > > --- a/arch/powerpc/mm/mmap.c > > +++ b/arch/powerpc/mm/mmap.c > > @@ -106,13 +106,17 @@ radix__arch_get_unmapped_area(struct file *filp, unsigned long addr, > > { > > struct mm_struct *mm = current->mm; > > struct vm_area_struct *vma; > > + unsigned long high_limit = DEFAULT_MAP_WINDOW; > > struct vm_unmapped_area_info info; > > > > if (unlikely(addr > mm->context.addr_limit && > > mm->context.addr_limit != TASK_SIZE)) > > mm->context.addr_limit = TASK_SIZE; > > > > - if (len > mm->task_size - mmap_min_addr) > > + if (addr > high_limit) > > + high_limit = TASK_SIZE; > > + > > + if (len > high_limit - mmap_min_addr) > > return -ENOMEM; > > > > if (flags & MAP_FIXED) > > @@ -121,7 +125,7 @@ radix__arch_get_unmapped_area(struct file *filp, unsigned long addr, > > if (addr) { > > addr = PAGE_ALIGN(addr); > > vma = find_vma(mm, addr); > > - if (mm->task_size - len >= addr && addr >= mmap_min_addr && > > + if (high_limit - len >= addr && addr >= mmap_min_addr && > > (!vma || addr + len <= vm_start_gap(vma))) > > return addr; > > } > > @@ -129,13 +133,9 @@ radix__arch_get_unmapped_area(struct file *filp, unsigned long addr, > > info.flags = 0; > > info.length = len; > > info.low_limit = mm->mmap_base; > > + info.high_limit = high_limit; > > info.align_mask = 0; > > > > - if (unlikely(addr > DEFAULT_MAP_WINDOW)) > > - info.high_limit = mm->context.addr_limit; > > - else > > - info.high_limit = DEFAULT_MAP_WINDOW; > > - > > return vm_unmapped_area(&info); > > } > > > > @@ -149,14 +149,18 @@ radix__arch_get_unmapped_area_topdown(struct file *filp, > > struct vm_area_struct *vma; > > struct mm_struct *mm = current->mm; > > unsigned long addr = addr0; > > + unsigned long high_limit = DEFAULT_MAP_WINDOW; > > struct vm_unmapped_area_info info; > > > > if (unlikely(addr > mm->context.addr_limit && > > mm->context.addr_limit != TASK_SIZE)) > > mm->context.addr_limit = TASK_SIZE; > > > > + if (addr > high_limit) > > + high_limit = TASK_SIZE; > > + > > /* requested length too big for entire address space */ > > - if (len > mm->task_size - mmap_min_addr) > > + if (len > high_limit - mmap_min_addr) > > return -ENOMEM; > > > > if (flags & MAP_FIXED) > > @@ -166,7 +170,7 @@ radix__arch_get_unmapped_area_topdown(struct file *filp, > > if (addr) { > > addr = PAGE_ALIGN(addr); > > vma = find_vma(mm, addr); > > - if (mm->task_size - len >= addr && addr >= mmap_min_addr && > > + if (high_limit - len >= addr && addr >= mmap_min_addr && > > (!vma || addr + len <= vm_start_gap(vma))) > > return addr; > > } > > @@ -174,12 +178,9 @@ radix__arch_get_unmapped_area_topdown(struct file *filp, > > info.flags = VM_UNMAPPED_AREA_TOPDOWN; > > info.length = len; > > info.low_limit = max(PAGE_SIZE, mmap_min_addr); > > - info.high_limit = mm->mmap_base; > > + info.high_limit = mm->mmap_base + (high_limit - DEFAULT_MAP_WINDOW); > > info.align_mask = 0; > > > > - if (addr > DEFAULT_MAP_WINDOW) > > - info.high_limit += mm->context.addr_limit - DEFAULT_MAP_WINDOW; > > - > > addr = vm_unmapped_area(&info); > > if (!(addr & ~PAGE_MASK)) > > return addr; > > -- > > 2.15.0 >