From: Vincenzo Frascino <vincenzo.frascino@arm.com>
To: Catalin Marinas <catalin.marinas@arm.com>,
Andrey Konovalov <andreyknvl@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>,
kasan-dev@googlegroups.com,
Andrey Ryabinin <aryabinin@virtuozzo.com>,
Alexander Potapenko <glider@google.com>,
Marco Elver <elver@google.com>,
Evgenii Stepanov <eugenis@google.com>,
Elena Petrova <lenaptr@google.com>,
Branislav Rankov <Branislav.Rankov@arm.com>,
Kevin Brodsky <kevin.brodsky@arm.com>,
Will Deacon <will.deacon@arm.com>,
Andrew Morton <akpm@linux-foundation.org>,
linux-arm-kernel@lists.infradead.org, linux-mm@kvack.org,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2 22/37] arm64: mte: Add in-kernel MTE helpers
Date: Thu, 17 Sep 2020 15:21:41 +0100 [thread overview]
Message-ID: <7904f7c2-cf3b-315f-8885-e8709c232718@arm.com> (raw)
In-Reply-To: <20200917134653.GB10662@gaia>
On 9/17/20 2:46 PM, Catalin Marinas wrote:
> On Tue, Sep 15, 2020 at 11:16:04PM +0200, Andrey Konovalov wrote:
>> diff --git a/arch/arm64/include/asm/mte-helpers.h b/arch/arm64/include/asm/mte-helpers.h
>> new file mode 100644
>> index 000000000000..5dc2d443851b
>> --- /dev/null
>> +++ b/arch/arm64/include/asm/mte-helpers.h
>> @@ -0,0 +1,48 @@
>> +/* SPDX-License-Identifier: GPL-2.0 */
>> +/*
>> + * Copyright (C) 2020 ARM Ltd.
>> + */
>> +#ifndef __ASM_MTE_ASM_H
>> +#define __ASM_MTE_ASM_H
>> +
>> +#define __MTE_PREAMBLE ".arch armv8.5-a\n.arch_extension memtag\n"
>
> Because of how the .arch overrides a previous .arch, we should follow
> the ARM64_ASM_PREAMBLE introduced in commit 1764c3edc668 ("arm64: use a
> common .arch preamble for inline assembly"). The above should be
> something like:
>
> #define __MTE_PREAMBLE ARM64_ASM_PREAMBLE ".arch_extension memtag"
>
> with the ARM64_ASM_PREAMBLE adjusted to armv8.5-a if available.
Good idea, I was not aware of commit 1764c3edc668. I will fix it accordingly.
>
>> +#define MTE_GRANULE_SIZE UL(16)
>> +#define MTE_GRANULE_MASK (~(MTE_GRANULE_SIZE - 1))
>> +#define MTE_TAG_SHIFT 56
>> +#define MTE_TAG_SIZE 4
>> +#define MTE_TAG_MASK GENMASK((MTE_TAG_SHIFT + (MTE_TAG_SIZE - 1)), MTE_TAG_SHIFT)
>> +#define MTE_TAG_MAX (MTE_TAG_MASK >> MTE_TAG_SHIFT)
>
> In v1 I suggested we keep those definitions in mte-def.h (or
> mte-hwdef.h) so that they can be included in cache.h. Anything else
> should go in mte.h, I don't see the point of two headers for various MTE
> function prototypes.
>
This is what I did in my patches I shared with Andrey. I suppose that since in
this version he introduced some functions that are needed in this file, he
reverted to the old name (mte-helper.h).
>> +
>> +#ifndef __ASSEMBLY__
>> +
>> +#include <linux/types.h>
>> +
>> +#ifdef CONFIG_ARM64_MTE
>> +
>> +#define mte_get_ptr_tag(ptr) ((u8)(((u64)(ptr)) >> MTE_TAG_SHIFT))
>
> I wonder whether this could also be an inline function that takes a void
> *ptr.
>
>> diff --git a/arch/arm64/kernel/mte.c b/arch/arm64/kernel/mte.c
>> index 52a0638ed967..e238ffde2679 100644
>> --- a/arch/arm64/kernel/mte.c
>> +++ b/arch/arm64/kernel/mte.c
>> @@ -72,6 +74,52 @@ int memcmp_pages(struct page *page1, struct page *page2)
>> return ret;
>> }
>>
>> +u8 mte_get_mem_tag(void *addr)
>> +{
>> + if (system_supports_mte())
>> + asm volatile(ALTERNATIVE("ldr %0, [%0]",
>> + __MTE_PREAMBLE "ldg %0, [%0]",
>> + ARM64_MTE)
>> + : "+r" (addr));
>
> This doesn't do what you think it does. LDG indeed reads the tag from
> memory but LDR loads the actual data at that address. Instead of the
> first LDR, you may want something like "mov %0, #0xf << 56" (and use
> some macros to avoid the hard-coded 56).
>
The result of the load should never be used since it is meaningful only if
system_supports_mte(). It should be only required for compilation purposes.
Said that, I think I like more your solution hence I am going to adopt it.
>> +
>> + return 0xF0 | mte_get_ptr_tag(addr);
>> +}
>> +
>> +u8 mte_get_random_tag(void)
>> +{
>> + u8 tag = 0xF;
>> + u64 addr = 0;
>> +
>> + if (system_supports_mte()) {
>> + asm volatile(ALTERNATIVE("add %0, %0, %0",
>> + __MTE_PREAMBLE "irg %0, %0",
>> + ARM64_MTE)
>> + : "+r" (addr));
>
> What was the intention here? The first ADD doubles the pointer value and
> gets a tag out of it (possibly doubled as well, depends on the carry
> from bit 55). Better use something like "orr %0, %0, #0xf << 56".
>
Same as above but I will use the orr in the next version.
>> +
>> + tag = mte_get_ptr_tag(addr);
>> + }
>> +
>> + return 0xF0 | tag;
>
> This function return seems inconsistent with the previous one. I'd
> prefer the return line to be the same in both.
>
The reason why it is different is that in this function extracting the tag from
the address makes sense only if irg is executed.
I can initialize addr to 0xf << 56 and make them the same.
>> +}
>> +
>> +void *mte_set_mem_tag_range(void *addr, size_t size, u8 tag)
>> +{
>> + void *ptr = addr;
>> +
>> + if ((!system_supports_mte()) || (size == 0))
>> + return addr;
>> +
>> + /* Make sure that size is aligned. */
>> + WARN_ON(size & (MTE_GRANULE_SIZE - 1));
>> +
>> + tag = 0xF0 | (tag & 0xF);
>
> No point in tag & 0xf, the top nibble doesn't matter as you or 0xf0 in.
>
Agree, will remove in the next version.
>> + ptr = (void *)__tag_set(ptr, tag);
>> +
>> + mte_assign_mem_tag_range(ptr, size);
>> +
>> + return ptr;
>> +}
>> +
>> static void update_sctlr_el1_tcf0(u64 tcf0)
>> {
>> /* ISB required for the kernel uaccess routines */
>> diff --git a/arch/arm64/lib/mte.S b/arch/arm64/lib/mte.S
>> index 03ca6d8b8670..cc2c3a378c00 100644
>> --- a/arch/arm64/lib/mte.S
>> +++ b/arch/arm64/lib/mte.S
>> @@ -149,3 +149,20 @@ SYM_FUNC_START(mte_restore_page_tags)
>>
>> ret
>> SYM_FUNC_END(mte_restore_page_tags)
>> +
>> +/*
>> + * Assign allocation tags for a region of memory based on the pointer tag
>> + * x0 - source pointer
>> + * x1 - size
>> + *
>> + * Note: size must be non-zero and MTE_GRANULE_SIZE aligned
>> + */
>> +SYM_FUNC_START(mte_assign_mem_tag_range)
>> + /* if (src == NULL) return; */
>> + cbz x0, 2f
>> +1: stg x0, [x0]
>> + add x0, x0, #MTE_GRANULE_SIZE
>> + sub x1, x1, #MTE_GRANULE_SIZE
>> + cbnz x1, 1b
>> +2: ret
>> +SYM_FUNC_END(mte_assign_mem_tag_range)
>
> I thought Vincenzo agreed to my comments on the previous version w.r.t.
> the fist cbz and the last cbnz:
>
> https://lore.kernel.org/linux-arm-kernel/921c4ed0-b5b5-bc01-5418-c52d80f1af59@arm.com/
>
Ups, this is my fault, I just realized I missed to unstash this change. Will be
present in the next version.
--
Regards,
Vincenzo
next prev parent reply other threads:[~2020-09-17 14:28 UTC|newest]
Thread overview: 90+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-09-15 21:15 [PATCH v2 00/37] kasan: add hardware tag-based mode for arm64 Andrey Konovalov
2020-09-15 21:15 ` [PATCH v2 01/37] kasan: KASAN_VMALLOC depends on KASAN_GENERIC Andrey Konovalov
2020-09-15 21:15 ` [PATCH v2 02/37] kasan: group vmalloc code Andrey Konovalov
2020-09-15 21:15 ` [PATCH v2 03/37] kasan: shadow declarations only for software modes Andrey Konovalov
2020-09-15 21:15 ` [PATCH v2 04/37] kasan: rename (un)poison_shadow to (un)poison_memory Andrey Konovalov
2020-09-15 21:15 ` [PATCH v2 05/37] kasan: rename KASAN_SHADOW_* to KASAN_GRANULE_* Andrey Konovalov
2020-09-18 8:04 ` Alexander Potapenko
2020-09-18 10:42 ` Andrey Konovalov
2020-09-15 21:15 ` [PATCH v2 06/37] kasan: only build init.c for software modes Andrey Konovalov
2020-09-15 21:15 ` [PATCH v2 07/37] kasan: split out shadow.c from common.c Andrey Konovalov
2020-09-18 8:17 ` Alexander Potapenko
2020-09-18 10:39 ` Andrey Konovalov
2020-09-15 21:15 ` [PATCH v2 08/37] kasan: rename generic/tags_report.c files Andrey Konovalov
2020-09-15 21:15 ` [PATCH v2 09/37] kasan: don't duplicate config dependencies Andrey Konovalov
2020-09-15 21:15 ` [PATCH v2 10/37] kasan: hide invalid free check implementation Andrey Konovalov
2020-09-15 21:15 ` [PATCH v2 11/37] kasan: decode stack frame only with KASAN_STACK_ENABLE Andrey Konovalov
2020-09-15 21:15 ` [PATCH v2 12/37] kasan, arm64: only init shadow for software modes Andrey Konovalov
2020-09-17 17:05 ` Catalin Marinas
2020-09-15 21:15 ` [PATCH v2 13/37] kasan, arm64: only use kasan_depth " Andrey Konovalov
2020-09-17 17:05 ` Catalin Marinas
2020-09-15 21:15 ` [PATCH v2 14/37] kasan: rename addr_has_shadow to addr_has_metadata Andrey Konovalov
2020-09-15 21:15 ` [PATCH v2 15/37] kasan: rename print_shadow_for_address to print_memory_metadata Andrey Konovalov
2020-09-15 21:15 ` [PATCH v2 16/37] kasan: kasan_non_canonical_hook only for software modes Andrey Konovalov
2020-09-15 21:15 ` [PATCH v2 17/37] kasan: rename SHADOW layout macros to META Andrey Konovalov
2020-09-15 21:16 ` [PATCH v2 18/37] kasan: separate metadata_fetch_row for each mode Andrey Konovalov
2020-09-15 21:16 ` [PATCH v2 19/37] kasan: don't allow SW_TAGS with ARM64_MTE Andrey Konovalov
2020-09-17 17:05 ` Catalin Marinas
2020-09-15 21:16 ` [PATCH v2 20/37] kasan: rename tags.c to tags_sw.c Andrey Konovalov
2020-09-18 9:41 ` Alexander Potapenko
2020-09-18 9:44 ` Alexander Potapenko
2020-09-18 9:46 ` Alexander Potapenko
2020-09-18 10:42 ` Andrey Konovalov
2020-09-15 21:16 ` [PATCH v2 21/37] kasan: introduce CONFIG_KASAN_HW_TAGS Andrey Konovalov
2020-09-18 12:32 ` Marco Elver
2020-09-18 15:06 ` Andrey Konovalov
2020-09-18 15:36 ` Marco Elver
2020-09-18 15:45 ` Andrey Konovalov
2020-09-15 21:16 ` [PATCH v2 22/37] arm64: mte: Add in-kernel MTE helpers Andrey Konovalov
2020-09-17 13:46 ` Catalin Marinas
2020-09-17 14:21 ` Vincenzo Frascino [this message]
2020-09-18 9:36 ` Catalin Marinas
2020-09-22 10:16 ` Vincenzo Frascino
2020-09-17 16:17 ` Vincenzo Frascino
2020-09-17 17:07 ` Catalin Marinas
2020-09-15 21:16 ` [PATCH v2 23/37] arm64: kasan: Add arch layer for memory tagging helpers Andrey Konovalov
2020-09-17 17:05 ` Catalin Marinas
2020-09-18 13:00 ` Marco Elver
2020-09-18 14:56 ` Andrey Konovalov
2020-09-15 21:16 ` [PATCH v2 24/37] arm64: mte: Add in-kernel tag fault handler Andrey Konovalov
2020-09-17 14:03 ` Catalin Marinas
2020-09-17 14:24 ` Vincenzo Frascino
2020-09-17 14:59 ` Catalin Marinas
2020-09-15 21:16 ` [PATCH v2 25/37] arm64: kasan: Enable in-kernel MTE Andrey Konovalov
2020-09-17 16:35 ` Catalin Marinas
2020-09-15 21:16 ` [PATCH v2 26/37] arm64: mte: Convert gcr_user into an exclude mask Andrey Konovalov
2020-09-17 17:06 ` Catalin Marinas
2020-09-15 21:16 ` [PATCH v2 27/37] arm64: mte: Switch GCR_EL1 in kernel entry and exit Andrey Konovalov
2020-09-17 16:52 ` Catalin Marinas
2020-09-17 16:58 ` Catalin Marinas
2020-09-17 18:47 ` Vincenzo Frascino
2020-09-18 9:39 ` Catalin Marinas
2020-09-15 21:16 ` [PATCH v2 28/37] arm64: kasan: Enable TBI EL1 Andrey Konovalov
2020-09-17 16:54 ` Catalin Marinas
2020-09-15 21:16 ` [PATCH v2 29/37] arm64: kasan: Align allocations for HW_TAGS Andrey Konovalov
2020-09-17 17:06 ` Catalin Marinas
2020-09-15 21:16 ` [PATCH v2 30/37] kasan: define KASAN_GRANULE_SIZE " Andrey Konovalov
2020-09-15 21:16 ` [PATCH v2 31/37] kasan, x86, s390: update undef CONFIG_KASAN Andrey Konovalov
2020-09-18 10:52 ` Marco Elver
2020-09-18 15:07 ` Andrey Konovalov
2020-09-24 21:35 ` Andrey Konovalov
2020-09-15 21:16 ` [PATCH v2 32/37] kasan, arm64: expand CONFIG_KASAN checks Andrey Konovalov
2020-09-17 17:06 ` Catalin Marinas
2020-09-15 21:16 ` [PATCH v2 33/37] kasan, arm64: implement HW_TAGS runtime Andrey Konovalov
2020-09-17 17:06 ` Catalin Marinas
2020-09-18 10:46 ` Marco Elver
2020-09-18 12:28 ` Andrey Konovalov
2020-09-18 12:52 ` Marco Elver
2020-09-18 15:00 ` Andrey Konovalov
2020-09-18 15:19 ` Marco Elver
2020-09-18 15:52 ` Andrey Konovalov
2020-09-15 21:16 ` [PATCH v2 34/37] kasan, arm64: print report from tag fault handler Andrey Konovalov
2020-09-17 17:04 ` Catalin Marinas
2020-09-18 12:26 ` Andrey Konovalov
2020-09-15 21:16 ` [PATCH v2 35/37] kasan, slub: reset tags when accessing metadata Andrey Konovalov
2020-09-18 14:44 ` Marco Elver
2020-09-18 14:55 ` Andrey Konovalov
2020-09-18 15:29 ` Catalin Marinas
2020-09-15 21:16 ` [PATCH v2 36/37] kasan, arm64: enable CONFIG_KASAN_HW_TAGS Andrey Konovalov
2020-09-17 17:04 ` Catalin Marinas
2020-09-15 21:16 ` [PATCH v2 37/37] kasan: add documentation for hardware tag-based mode Andrey Konovalov
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=7904f7c2-cf3b-315f-8885-e8709c232718@arm.com \
--to=vincenzo.frascino@arm.com \
--cc=Branislav.Rankov@arm.com \
--cc=akpm@linux-foundation.org \
--cc=andreyknvl@google.com \
--cc=aryabinin@virtuozzo.com \
--cc=catalin.marinas@arm.com \
--cc=dvyukov@google.com \
--cc=elver@google.com \
--cc=eugenis@google.com \
--cc=glider@google.com \
--cc=kasan-dev@googlegroups.com \
--cc=kevin.brodsky@arm.com \
--cc=lenaptr@google.com \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=will.deacon@arm.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).