From: "Ard Biesheuvel" <ard.biesheuvel@linaro.org>
To: "Philippe Mathieu-Daudé" <philmd@redhat.com>
Cc: edk2-devel-groups-io <devel@edk2.groups.io>,
Leif Lindholm <leif.lindholm@linaro.org>,
Laszlo Ersek <lersek@redhat.com>
Subject: Re: [PATCH v2] ArmVirtPkg: increase FD/FV size for NOOPT builds
Date: Thu, 12 Sep 2019 18:13:18 +0100 [thread overview]
Message-ID: <CAKv+Gu9Gy0ceQvRBdQiJZ8ekYoneiZP1fv8_uH73pxU+U=kimQ@mail.gmail.com> (raw)
In-Reply-To: <5141ec32-a51d-3ec6-79ef-ceec43b32c84@redhat.com>
On Thu, 12 Sep 2019 at 18:05, Philippe Mathieu-Daudé <philmd@redhat.com> wrote:
>
> On 9/12/19 6:50 PM, Ard Biesheuvel wrote:
> > On Thu, 12 Sep 2019 at 16:20, Philippe Mathieu-Daudé <philmd@redhat.com> wrote:
> >>
> >> Hi Ard,
> >>
> >> On 9/11/19 6:23 PM, Ard Biesheuvel wrote:
> >>> After upgrading the CI system we use for building the ArmVirtPkg
> >>> targets, we started seeing failures due to the NOOPT build running
> >>> out of space when using the CLANG38 toolchain definition combined
> >>> with clang 7.
> >>>
> >>> We really don't want to increase the FD/FV sizes in general to
> >>> accommodate this, so parameterize the relevant quantities and
> >>> increase them by 50% for NOOPT builds.
> >>>
> >>> Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
> >>> ---
> >>> v2: implement suggestions by Laszlo on 1) how to parameterize this further,
> >>> and b) to avoid adding another .inc file
> >>> update kernel header field, as pointed out by Philippe
> >>>
> >>> ArmVirtPkg/ArmVirt.dsc.inc | 28 ++++++++++++++++++++
> >>> ArmVirtPkg/ArmVirtQemu.fdf | 14 +++++++---
> >>> ArmVirtPkg/ArmVirtQemuKernel.fdf | 19 ++++++++++---
> >>> ArmVirtPkg/ArmVirtXen.fdf | 14 +++++++---
> >>> 4 files changed, 66 insertions(+), 9 deletions(-)
> >>>
> >>> diff --git a/ArmVirtPkg/ArmVirt.dsc.inc b/ArmVirtPkg/ArmVirt.dsc.inc
> >>> index a4ae25d982a2..d6b58e5c018b 100644
> >>> --- a/ArmVirtPkg/ArmVirt.dsc.inc
> >>> +++ b/ArmVirtPkg/ArmVirt.dsc.inc
> >>> @@ -10,6 +10,34 @@
> >>> [Defines]
> >>> DEFINE DEBUG_PRINT_ERROR_LEVEL = 0x8000004F
> >>>
> >>> + #
> >>> + # Flash size selection. Setting FD_SIZE_IN_KB on the command line directly to
> >>> + # one of the supported values, in place of any of the convenience macros, is
> >>> + # permitted.
> >>> + #
> >>> +!if $(TARGET) == NOOPT
> >>> + DEFINE FD_SIZE_3MB = TRUE
> >>> +!endif
> >>> +
> >>> +!ifdef $(FD_SIZE_2MB)
> >>> + DEFINE FD_SIZE_IN_KB = 2048
> >>> +!else
> >>> +!ifdef $(FD_SIZE_3MB)
> >>> + DEFINE FD_SIZE_IN_KB = 3072
> >>> +!else
> >>> + DEFINE FD_SIZE_IN_KB = 2048
> >>> +!endif
> >>> +!endif
> >>> +
> >>> +!if $(FD_SIZE_IN_KB) == 2048
> >>> + DEFINE FD_SIZE = 0x200000
> >>> + DEFINE FD_NUM_BLOCKS = 0x200
> >>> +!endif
> >>> +!if $(FD_SIZE_IN_KB) == 3072
> >>> + DEFINE FD_SIZE = 0x300000
> >>> + DEFINE FD_NUM_BLOCKS = 0x300
> >>> +!endif
> >>> +
> >>> [BuildOptions.common.EDKII.DXE_CORE,BuildOptions.common.EDKII.DXE_DRIVER,BuildOptions.common.EDKII.UEFI_DRIVER,BuildOptions.common.EDKII.UEFI_APPLICATION]
> >>> GCC:*_*_*_DLINK_FLAGS = -z common-page-size=0x1000
> >>>
> >>> diff --git a/ArmVirtPkg/ArmVirtQemu.fdf b/ArmVirtPkg/ArmVirtQemu.fdf
> >>> index c2169cb7964b..d3950c8be05e 100644
> >>> --- a/ArmVirtPkg/ArmVirtQemu.fdf
> >>> +++ b/ArmVirtPkg/ArmVirtQemu.fdf
> >>> @@ -20,14 +20,22 @@
> >>> #
> >>> ################################################################################
> >>>
> >>> +[Defines]
> >>> +!if $(FD_SIZE_IN_KB) == 2048
> >>> + DEFINE FVMAIN_COMPACT_SIZE = 0x1ff000
> >>> +!endif
> >>> +!if $(FD_SIZE_IN_KB) == 3072
> >>> + DEFINE FVMAIN_COMPACT_SIZE = 0x2ff000
> >>> +!endif
> >>> +
> >>> [FD.QEMU_EFI]
> >>> BaseAddress = 0x00000000|gArmTokenSpaceGuid.PcdFdBaseAddress # QEMU assigns 0 - 0x8000000 for a BootROM
> >>> -Size = 0x00200000|gArmTokenSpaceGuid.PcdFdSize # The size in bytes of the FLASH Device
> >>> +Size = $(FD_SIZE)|gArmTokenSpaceGuid.PcdFdSize # The size in bytes of the FLASH Device
> >>> ErasePolarity = 1
> >>>
> >>> # This one is tricky, it must be: BlockSize * NumBlocks = Size
> >>> BlockSize = 0x00001000
> >>> -NumBlocks = 0x200
> >>> +NumBlocks = $(FD_NUM_BLOCKS)
> >>>
> >>> ################################################################################
> >>> #
> >>> @@ -59,7 +67,7 @@ DATA = {
> >>> !endif
> >>> }
> >>>
> >>> -0x00001000|0x001ff000
> >>> +0x00001000|$(FVMAIN_COMPACT_SIZE)
> >>> gArmTokenSpaceGuid.PcdFvBaseAddress|gArmTokenSpaceGuid.PcdFvSize
> >>> FV = FVMAIN_COMPACT
> >>>
> >>> diff --git a/ArmVirtPkg/ArmVirtQemuKernel.fdf b/ArmVirtPkg/ArmVirtQemuKernel.fdf
> >>> index f675b6d65ee1..46ec967e1cc0 100644
> >>> --- a/ArmVirtPkg/ArmVirtQemuKernel.fdf
> >>> +++ b/ArmVirtPkg/ArmVirtQemuKernel.fdf
> >>> @@ -20,14 +20,22 @@
> >>> #
> >>> ################################################################################
> >>>
> >>> +[Defines]
> >>> +!if $(FD_SIZE_IN_KB) == 2048
> >>> + DEFINE FVMAIN_COMPACT_SIZE = 0x1f8000
> >>> +!endif
> >>> +!if $(FD_SIZE_IN_KB) == 3072
> >>> + DEFINE FVMAIN_COMPACT_SIZE = 0x2f8000
> >>> +!endif
> >>> +
> >>> [FD.QEMU_EFI]
> >>> BaseAddress = 0x00000000|gArmTokenSpaceGuid.PcdFdBaseAddress # QEMU assigns 0 - 0x8000000 for a BootROM
> >>> -Size = 0x00200000|gArmTokenSpaceGuid.PcdFdSize # The size in bytes of the FLASH Device
> >>> +Size = $(FD_SIZE)|gArmTokenSpaceGuid.PcdFdSize # The size in bytes of the FLASH Device
> >>> ErasePolarity = 1
> >>>
> >>> # This one is tricky, it must be: BlockSize * NumBlocks = Size
> >>> BlockSize = 0x00001000
> >>> -NumBlocks = 0x200
> >>> +NumBlocks = $(FD_NUM_BLOCKS)
> >>>
> >>> ################################################################################
> >>> #
> >>> @@ -56,7 +64,12 @@ DATA = {
> >>> 0x01, 0x00, 0x00, 0x10, # code0: adr x1, .
> >>> 0xff, 0x1f, 0x00, 0x14, # code1: b 0x8000
> >>> 0x00, 0x00, 0x08, 0x00, 0x00, 0x00, 0x00, 0x00, # text_offset: 512 KB
> >>> +!if $(FD_SIZE_IN_KB) == 2048
> >>> 0x00, 0x00, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, # image_size: 2 MB
> >>> +!endif
> >>> +!if $(FD_SIZE_IN_KB) == 3072
> >>> + 0x00, 0x00, 0x30, 0x00, 0x00, 0x00, 0x00, 0x00, # image_size: 3 MB
> >>> +!endif
> >>
> >> Ah, I was thinking of some inplace endian swapping
> >>
> >
> > If you can make it work, patches welcome :-)
>
> I was thinking of this:
>
> 0xf6, 0x1f, 0x00, 0xea, # b 0x8000
>
> 0x18, 0x28, 0x6f, 0x01, # magic
>
> 0x00, 0x00, 0x00, 0x00, # start
>
> ((($(FD_SIZE)) & 0x000000ff) << 24), # image size
>
> ((($(FD_SIZE)) & 0x0000ff00) << 8),
>
> ((($(FD_SIZE)) & 0x00ff0000) >> 8),
>
> ((($(FD_SIZE)) & 0xff000000) >> 24),
>
> 0x01, 0x02, 0x03, 0x04 # endiannness flag
>
>
> But now that I look at it, I realize it is uglier to review than your
> if/endif...
>
Does the .fdf parser even support that kind of arithmetic? :-)
next prev parent reply other threads:[~2019-09-12 17:13 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-09-11 16:23 [PATCH v2] ArmVirtPkg: increase FD/FV size for NOOPT builds Ard Biesheuvel
2019-09-12 14:46 ` Laszlo Ersek
2019-09-12 14:47 ` Laszlo Ersek
2019-09-12 14:53 ` Ard Biesheuvel
2019-09-12 15:20 ` Philippe Mathieu-Daudé
2019-09-12 16:50 ` Ard Biesheuvel
2019-09-12 17:05 ` Philippe Mathieu-Daudé
2019-09-12 17:13 ` Ard Biesheuvel [this message]
2019-09-12 17:14 ` Philippe Mathieu-Daudé
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-list 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+Gu9Gy0ceQvRBdQiJZ8ekYoneiZP1fv8_uH73pxU+U=kimQ@mail.gmail.com' \
--to=devel@edk2.groups.io \
/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