linux-kernel - Re: [PATCH V9 1/4] perf/core: Add PERF_SAMPLE_DATA_PAGE

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <877drz1qbc.fsf@mpe.ellerman.id.au>
Date:   Sat, 10 Oct 2020 00:28:39 +1100
From:   Michael Ellerman <mpe@...erman.id.au>
To:     Peter Zijlstra <peterz@...radead.org>, kan.liang@...ux.intel.com
Cc:     mingo@...hat.com, acme@...nel.org, linux-kernel@...r.kernel.org,
        mark.rutland@....com, alexander.shishkin@...ux.intel.com,
        jolsa@...hat.com, eranian@...gle.com, ak@...ux.intel.com,
        dave.hansen@...el.com, kirill.shutemov@...ux.intel.com,
        benh@...nel.crashing.org, paulus@...ba.org,
        Will Deacon <will@...nel.org>,
        David Miller <davem@...emloft.net>
Subject: Re: [PATCH V9 1/4] perf/core: Add PERF_SAMPLE_DATA_PAGE_SIZE

Peter Zijlstra <peterz@...radead.org> writes:
> On Thu, Oct 01, 2020 at 06:57:46AM -0700, kan.liang@...ux.intel.com wrote:
>> +/*
>> + * Return the MMU page size of a given virtual address
>> + */
>> +static u64 __perf_get_page_size(struct mm_struct *mm, unsigned long addr)
>> +{
>> +	pgd_t *pgd;
>> +	p4d_t *p4d;
>> +	pud_t *pud;
>> +	pmd_t *pmd;
>> +	pte_t *pte;
>> +
>> +	pgd = pgd_offset(mm, addr);
>> +	if (pgd_none(*pgd))
>> +		return 0;
>> +
>> +	p4d = p4d_offset(pgd, addr);
>> +	if (!p4d_present(*p4d))
>> +		return 0;
>> +
>> +	if (p4d_leaf(*p4d))
>> +		return 1ULL << P4D_SHIFT;
>> +
>> +	pud = pud_offset(p4d, addr);
>> +	if (!pud_present(*pud))
>> +		return 0;
>> +
>> +	if (pud_leaf(*pud))
>> +		return 1ULL << PUD_SHIFT;
>> +
>> +	pmd = pmd_offset(pud, addr);
>> +	if (!pmd_present(*pmd))
>> +		return 0;
>> +
>> +	if (pmd_leaf(*pmd))
>> +		return 1ULL << PMD_SHIFT;
>> +
>> +	pte = pte_offset_map(pmd, addr);
>> +	if (!pte_present(*pte)) {
>> +		pte_unmap(pte);
>> +		return 0;
>> +	}
>> +
>> +	pte_unmap(pte);
>> +	return PAGE_SIZE;
>> +}
>
> So this mostly works, but gets a number of hugetlb and arch specific
> things wrong.
>
> With the first 3 patches, this is only exposed to x86 and Power.
> Michael, does the above work for you?

It might work on our server CPUs (book3s) but in general no I don't
think it will work for all powerpc.

> Looking at:
>
> arch/powerpc/include/asm/book3s/64/hugetlb.h:check_and_get_huge_psize()
>
> You seem to limit yourself to page-table sizes, however if I then look
> at the same function in:
>
> arch/powerpc/include/asm/nohash/32/hugetlb-8xx.h
> arch/powerpc/include/asm/nohash/hugetlb-book3e.h
>
> it doesn't seem to constrain itself so.

Yeah the embedded CPUs have more weird sizes.

I think we have all the logic in our __find_linux_pte().

> Patch 4 makes it all far worse by exposing it to pretty much everybody.
>
> Now, I think we can fix at least the user mappings with the below delta,
> but if archs are using non-page-table MMU sizes we'll need arch helpers.
>
> ARM64 is in that last boat.

I think we probably need it to be weak so we can provide our own
version.

cheers