From: Nicholas Piggin <npiggin@gmail.com>
To: Christoph Hellwig <hch@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>,
Christophe Leroy <christophe.leroy@csgroup.eu>,
Ding Tianhong <dingtianhong@huawei.com>,
Jonathan Cameron <Jonathan.Cameron@huawei.com>,
linux-arch@vger.kernel.org, linux-kernel@vger.kernel.org,
linux-mm@kvack.org, linuxppc-dev@lists.ozlabs.org,
Zefan Li <lizefan@huawei.com>,
Rick Edgecombe <rick.p.edgecombe@intel.com>,
Randy Dunlap <rdunlap@infradead.org>
Subject: Re: [PATCH v10 11/12] mm/vmalloc: Hugepage vmalloc mappings
Date: Mon, 25 Jan 2021 09:17:18 +1000 [thread overview]
Message-ID: <1611529781.hxjbuadzrl.astroid@bobo.none> (raw)
In-Reply-To: <20210124150729.GC733865@infradead.org>
Excerpts from Christoph Hellwig's message of January 25, 2021 1:07 am:
> On Sun, Jan 24, 2021 at 06:22:29PM +1000, Nicholas Piggin wrote:
>> diff --git a/arch/Kconfig b/arch/Kconfig
>> index 24862d15f3a3..f87feb616184 100644
>> --- a/arch/Kconfig
>> +++ b/arch/Kconfig
>> @@ -724,6 +724,16 @@ config HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
>> config HAVE_ARCH_HUGE_VMAP
>> bool
>>
>> +config HAVE_ARCH_HUGE_VMALLOC
>> + depends on HAVE_ARCH_HUGE_VMAP
>> + bool
>> + help
>> + Archs that select this would be capable of PMD-sized vmaps (i.e.,
>> + arch_vmap_pmd_supported() returns true), and they must make no
>> + assumptions that vmalloc memory is mapped with PAGE_SIZE ptes. The
>> + VM_NOHUGE flag can be used to prohibit arch-specific allocations from
>> + using hugepages to help with this (e.g., modules may require it).
>
> help texts don't make sense for options that aren't user visible.
Yeah it was supposed to just be a comment but if it was user visible
then similar kind of thing would not make sense in help text, so I'll
just turn it into a real comment as per Randy's suggestion.
> More importantly, is there any good reason to keep the option and not
> just go the extra step and enable huge page vmalloc for arm64 and x86
> as well?
Yes they need to ensure they exclude vmallocs that can't be huge one
way or another (VM_ flag or prot arg).
After they're converted we can fold it into HUGE_VMAP.
>> +static inline bool is_vm_area_hugepages(const void *addr)
>> +{
>> + /*
>> + * This may not 100% tell if the area is mapped with > PAGE_SIZE
>> + * page table entries, if for some reason the architecture indicates
>> + * larger sizes are available but decides not to use them, nothing
>> + * prevents that. This only indicates the size of the physical page
>> + * allocated in the vmalloc layer.
>> + */
>> + return (find_vm_area(addr)->page_order > 0);
>
> No need for the braces here.
>
>> }
>>
>> +static int vmap_pages_range_noflush(unsigned long addr, unsigned long end,
>> + pgprot_t prot, struct page **pages, unsigned int page_shift)
>> +{
>> + unsigned int i, nr = (end - addr) >> PAGE_SHIFT;
>> +
>> + WARN_ON(page_shift < PAGE_SHIFT);
>> +
>> + if (page_shift == PAGE_SHIFT)
>> + return vmap_small_pages_range_noflush(addr, end, prot, pages);
>
> This begs for a IS_ENABLED check to disable the hugepage code for
> architectures that don't need it.
Yeah good point.
>> +int map_kernel_range_noflush(unsigned long addr, unsigned long size,
>> + pgprot_t prot, struct page **pages)
>> +{
>> + return vmap_pages_range_noflush(addr, addr + size, prot, pages, PAGE_SHIFT);
>> +}
>
> Please just kill off map_kernel_range_noflush and map_kernel_range
> off entirely in favor of the vmap versions.
I can do a cleanup patch on top of it.
>> + for (i = 0; i < area->nr_pages; i += 1U << area->page_order) {
>
> Maybe using a helper that takes the vm_area_struct and either returns
> area->page_order or always 0 based on IS_ENABLED?
I'll see how it looks.
Thanks,
Nick
next prev parent reply other threads:[~2021-01-24 23:18 UTC|newest]
Thread overview: 34+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-01-24 8:22 [PATCH v10 00/12] huge vmalloc mappings Nicholas Piggin
2021-01-24 8:22 ` [PATCH v10 01/12] mm/vmalloc: fix vmalloc_to_page for huge vmap mappings Nicholas Piggin
2021-01-24 11:31 ` Christoph Hellwig
2021-01-24 8:22 ` [PATCH v10 02/12] mm: apply_to_pte_range warn and fail if a large pte is encountered Nicholas Piggin
2021-01-24 11:32 ` Christoph Hellwig
2021-01-24 8:22 ` [PATCH v10 03/12] mm/vmalloc: rename vmap_*_range vmap_pages_*_range Nicholas Piggin
2021-01-24 11:34 ` Christoph Hellwig
2021-01-24 8:22 ` [PATCH v10 04/12] mm/ioremap: rename ioremap_*_range to vmap_*_range Nicholas Piggin
2021-01-24 11:36 ` Christoph Hellwig
2021-01-24 12:04 ` Nicholas Piggin
2021-01-24 8:22 ` [PATCH v10 05/12] mm: HUGE_VMAP arch support cleanup Nicholas Piggin
2021-01-24 11:40 ` Christoph Hellwig
2021-01-24 12:22 ` Nicholas Piggin
2021-01-25 8:19 ` Christophe Leroy
2021-01-25 8:40 ` Christophe Leroy
2021-01-24 8:22 ` [PATCH v10 06/12] powerpc: inline huge vmap supported functions Nicholas Piggin
2021-01-25 8:42 ` Christophe Leroy
2021-01-25 11:37 ` Nicholas Piggin
2021-01-24 8:22 ` [PATCH v10 07/12] arm64: " Nicholas Piggin
2021-01-24 8:22 ` [PATCH v10 08/12] x86: " Nicholas Piggin
2021-01-24 8:22 ` [PATCH v10 09/12] mm: Move vmap_range from mm/ioremap.c to mm/vmalloc.c Nicholas Piggin
2021-01-24 14:49 ` Christoph Hellwig
2021-01-24 8:22 ` [PATCH v10 10/12] mm/vmalloc: add vmap_range_noflush variant Nicholas Piggin
2021-01-24 14:51 ` Christoph Hellwig
2021-01-24 8:22 ` [PATCH v10 11/12] mm/vmalloc: Hugepage vmalloc mappings Nicholas Piggin
2021-01-24 15:07 ` Christoph Hellwig
2021-01-24 18:06 ` Randy Dunlap
2021-01-24 23:17 ` Nicholas Piggin [this message]
2021-01-25 9:14 ` Christophe Leroy
2021-01-25 11:37 ` Nicholas Piggin
2021-01-25 12:13 ` Christophe Leroy
2021-01-25 12:24 ` David Laight
2021-01-26 9:50 ` Nicholas Piggin
2021-01-24 8:22 ` [PATCH v10 12/12] powerpc/64s/radix: Enable huge " Nicholas Piggin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1611529781.hxjbuadzrl.astroid@bobo.none \
--to=npiggin@gmail.com \
--cc=Jonathan.Cameron@huawei.com \
--cc=akpm@linux-foundation.org \
--cc=christophe.leroy@csgroup.eu \
--cc=dingtianhong@huawei.com \
--cc=hch@infradead.org \
--cc=linux-arch@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=linuxppc-dev@lists.ozlabs.org \
--cc=lizefan@huawei.com \
--cc=rdunlap@infradead.org \
--cc=rick.p.edgecombe@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).