From: Ard Biesheuvel <ard.biesheuvel-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
To: Roy Franz <roy.franz-ZPxbGqLxI0U@public.gmane.org>
Cc: Matt Fleming
<matt-mF/unelCI9GS6iBeEJttW/XRex20P6io@public.gmane.org>,
"linux-efi-u79uwXL29TY76Z2rM5mHXA@public.gmane.org"
<linux-efi-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>,
Roy Franz <roy.franz-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
Subject: Re: [PATCH v3] Fix allocation size calculations
Date: Tue, 25 Oct 2016 19:03:00 +0100 [thread overview]
Message-ID: <CAKv+Gu8FDUQDD4-QCeB7ELFR9LHqr_qTHmjDVgp5xa6LnxNZXw@mail.gmail.com> (raw)
In-Reply-To: <20161025163634.11893-1-roy.franz-ZPxbGqLxI0U@public.gmane.org>
On 25 October 2016 at 17:36, Roy Franz <roy.franz-ZPxbGqLxI0U@public.gmane.org> wrote:
> Adjust the size used in calculations to match the actual size of allocation
> that will be performed based on EFI size/alignment constraints.
> efi_high_alloc() and efi_low_alloc() use the passed size in bytes directly
> to find space in the memory map for the allocation, rather than the actual
> allocation size that has been adjusted for size and alignment constraints.
> This results in failed allocations and retries in efi_high_alloc(). The
> same error is present in efi_low_alloc(), although failure will only happen
> if the lowest memory block is small.
> Also use EFI_PAGE_SIZE consistently and remove use of EFI_PAGE_SHIFT to
> calculate page size.
>
> Signed-off-by: Roy Franz <roy.franz-ZPxbGqLxI0U@public.gmane.org>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
> ---
>
> v3:
> * Revert incorrect additional size alignment
> * Simplify size and nr_pages calculation
> * Reword comments regarding alignment. Old comments mention page
> alignment, but for arm64 we enforce larger alignment. Clarify that
> the greater alignment is a Linux requirement, unlike the page alignment
> that is an EFI requirement.
> v2:
> * Add addition size alignment
> * Update comments regarding page alignment
>
>
> drivers/firmware/efi/libstub/efi-stub-helper.c | 24 ++++++++++++++----------
> 1 file changed, 14 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/firmware/efi/libstub/efi-stub-helper.c b/drivers/firmware/efi/libstub/efi-stub-helper.c
> index aded106..4b74bf8 100644
> --- a/drivers/firmware/efi/libstub/efi-stub-helper.c
> +++ b/drivers/firmware/efi/libstub/efi-stub-helper.c
> @@ -186,14 +186,16 @@ efi_status_t efi_high_alloc(efi_system_table_t *sys_table_arg,
> goto fail;
>
> /*
> - * Enforce minimum alignment that EFI requires when requesting
> - * a specific address. We are doing page-based allocations,
> - * so we must be aligned to a page.
> + * Enforce minimum alignment that EFI or Linux requires when
> + * requesting a specific address. We are doing page-based (or
> + * larger) allocations, and both the address and size must meet
> + * alignment constraints.
> */
> if (align < EFI_ALLOC_ALIGN)
> align = EFI_ALLOC_ALIGN;
>
> - nr_pages = round_up(size, EFI_ALLOC_ALIGN) / EFI_PAGE_SIZE;
> + size = round_up(size, EFI_ALLOC_ALIGN);
> + nr_pages = size / EFI_PAGE_SIZE;
> again:
> for (i = 0; i < map_size / desc_size; i++) {
> efi_memory_desc_t *desc;
> @@ -208,7 +210,7 @@ efi_status_t efi_high_alloc(efi_system_table_t *sys_table_arg,
> continue;
>
> start = desc->phys_addr;
> - end = start + desc->num_pages * (1UL << EFI_PAGE_SHIFT);
> + end = start + desc->num_pages * EFI_PAGE_SIZE;
>
> if (end > max)
> end = max;
> @@ -278,14 +280,16 @@ efi_status_t efi_low_alloc(efi_system_table_t *sys_table_arg,
> goto fail;
>
> /*
> - * Enforce minimum alignment that EFI requires when requesting
> - * a specific address. We are doing page-based allocations,
> - * so we must be aligned to a page.
> + * Enforce minimum alignment that EFI or Linux requires when
> + * requesting a specific address. We are doing page-based (or
> + * larger) allocations, and both the address and size must meet
> + * alignment constraints.
> */
> if (align < EFI_ALLOC_ALIGN)
> align = EFI_ALLOC_ALIGN;
>
> - nr_pages = round_up(size, EFI_ALLOC_ALIGN) / EFI_PAGE_SIZE;
> + size = round_up(size, EFI_ALLOC_ALIGN);
> + nr_pages = size / EFI_PAGE_SIZE;
> for (i = 0; i < map_size / desc_size; i++) {
> efi_memory_desc_t *desc;
> unsigned long m = (unsigned long)map;
> @@ -300,7 +304,7 @@ efi_status_t efi_low_alloc(efi_system_table_t *sys_table_arg,
> continue;
>
> start = desc->phys_addr;
> - end = start + desc->num_pages * (1UL << EFI_PAGE_SHIFT);
> + end = start + desc->num_pages * EFI_PAGE_SIZE;
>
> /*
> * Don't allocate at 0x0. It will confuse code that
> --
> 2.9.3
>
next prev parent reply other threads:[~2016-10-25 18:03 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-10-25 16:36 [PATCH v3] Fix allocation size calculations Roy Franz
[not found] ` <20161025163634.11893-1-roy.franz-ZPxbGqLxI0U@public.gmane.org>
2016-10-25 18:03 ` Ard Biesheuvel [this message]
[not found] ` <CAKv+Gu8FDUQDD4-QCeB7ELFR9LHqr_qTHmjDVgp5xa6LnxNZXw-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2016-10-31 7:57 ` Ard Biesheuvel
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=CAKv+Gu8FDUQDD4-QCeB7ELFR9LHqr_qTHmjDVgp5xa6LnxNZXw@mail.gmail.com \
--to=ard.biesheuvel-qsej5fyqhm4dnm+yrofe0a@public.gmane.org \
--cc=linux-efi-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
--cc=matt-mF/unelCI9GS6iBeEJttW/XRex20P6io@public.gmane.org \
--cc=roy.franz-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org \
--cc=roy.franz-ZPxbGqLxI0U@public.gmane.org \
/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).