From: "Laszlo Ersek" <lersek@redhat.com>
To: devel@edk2.groups.io, kraxel@redhat.com
Cc: Oliver Steffen <osteffen@redhat.com>,
Rahul Kumar <rahul1.kumar@intel.com>, Ray Ni <ray.ni@intel.com>
Subject: Re: [edk2-devel] [PATCH v2 5/5] UefiCpuPkg/MpInitLib: Add support for multiple HOBs to SaveCpuMpData()
Date: Thu, 22 Feb 2024 03:03:49 +0100 [thread overview]
Message-ID: <6ebafaae-25c6-e5c1-1c40-779cff5bfc24@redhat.com> (raw)
In-Reply-To: <20240220174939.1288689-6-kraxel@redhat.com>
On 2/20/24 18:49, Gerd Hoffmann wrote:
> Add support for splitting Hand-Off data into multiple HOBs. This is
> required for VMs with thousands of CPUs. The actual CPU count per HOB
> is much smaller (128) for better test coverage.
(1) The mention of the count 128 now seems stale for the code.
>
> Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
> ---
> UefiCpuPkg/Library/MpInitLib/PeiMpLib.c | 44 +++++++++++++++----------
> 1 file changed, 27 insertions(+), 17 deletions(-)
>
> diff --git a/UefiCpuPkg/Library/MpInitLib/PeiMpLib.c b/UefiCpuPkg/Library/MpInitLib/PeiMpLib.c
> index f80e00edcff3..8a916a218016 100644
> --- a/UefiCpuPkg/Library/MpInitLib/PeiMpLib.c
> +++ b/UefiCpuPkg/Library/MpInitLib/PeiMpLib.c
> @@ -126,35 +126,45 @@ SaveCpuMpData (
> IN CPU_MP_DATA *CpuMpData
> )
> {
> + UINT32 MaxCpusPerHob, CpusInHob;
> UINT64 Data64;
> - UINTN Index;
> + UINT32 Index, HobBase;
> CPU_INFO_IN_HOB *CpuInfoInHob;
> MP_HAND_OFF *MpHandOff;
> UINTN MpHandOffSize;
>
> + MaxCpusPerHob = (MAX_UINT16 - sizeof (EFI_HOB_GUID_TYPE) - sizeof (MP_HAND_OFF)) / sizeof (PROCESSOR_HAND_OFF);
(2) MAX_UINT16 should be 0xFFF8 instead; see subthread.
> +
> //
> // When APs are in a state that can be waken up by a store operation to a memory address,
> // report the MP_HAND_OFF data for DXE to use.
> //
> - CpuInfoInHob = (CPU_INFO_IN_HOB *)(UINTN)CpuMpData->CpuInfoInHob;
> - MpHandOffSize = sizeof (MP_HAND_OFF) + sizeof (PROCESSOR_HAND_OFF) * CpuMpData->CpuCount;
> - MpHandOff = (MP_HAND_OFF *)BuildGuidHob (&mMpHandOffGuid, MpHandOffSize);
> - ASSERT (MpHandOff != NULL);
> - ZeroMem (MpHandOff, MpHandOffSize);
> - MpHandOff->ProcessorIndex = 0;
> + CpuInfoInHob = (CPU_INFO_IN_HOB *)(UINTN)CpuMpData->CpuInfoInHob;
>
> - MpHandOff->CpuCount = CpuMpData->CpuCount;
> - if (CpuMpData->ApLoopMode != ApInHltLoop) {
> - MpHandOff->StartupSignalValue = MP_HAND_OFF_SIGNAL;
> - MpHandOff->WaitLoopExecutionMode = sizeof (VOID *);
> - }
> + for (Index = 0; Index < CpuMpData->CpuCount; Index++) {
> + if (Index % MaxCpusPerHob == 0) {
> + HobBase = Index;
> + CpusInHob = MIN (CpuMpData->CpuCount - HobBase, MaxCpusPerHob);
>
> - for (Index = 0; Index < MpHandOff->CpuCount; Index++) {
> - MpHandOff->Info[Index].ApicId = CpuInfoInHob[Index].ApicId;
> - MpHandOff->Info[Index].Health = CpuInfoInHob[Index].Health;
> + MpHandOffSize = sizeof (MP_HAND_OFF) + sizeof (PROCESSOR_HAND_OFF) * CpusInHob;
> + MpHandOff = (MP_HAND_OFF *)BuildGuidHob (&mMpHandOffGuid, MpHandOffSize);
> + ASSERT (MpHandOff != NULL);
> + ZeroMem (MpHandOff, MpHandOffSize);
> +
> + MpHandOff->ProcessorIndex = HobBase;
> + MpHandOff->CpuCount = CpusInHob;
> +
> + if (CpuMpData->ApLoopMode != ApInHltLoop) {
> + MpHandOff->StartupSignalValue = MP_HAND_OFF_SIGNAL;
> + MpHandOff->WaitLoopExecutionMode = sizeof (VOID *);
> + }
> + }
> +
> + MpHandOff->Info[Index-HobBase].ApicId = CpuInfoInHob[Index].ApicId;
> + MpHandOff->Info[Index-HobBase].Health = CpuInfoInHob[Index].Health;
> if (CpuMpData->ApLoopMode != ApInHltLoop) {
> - MpHandOff->Info[Index].StartupSignalAddress = (UINT64)(UINTN)CpuMpData->CpuData[Index].StartupApSignal;
> - MpHandOff->Info[Index].StartupProcedureAddress = (UINT64)(UINTN)&CpuMpData->CpuData[Index].ApFunction;
> + MpHandOff->Info[Index-HobBase].StartupSignalAddress = (UINT64)(UINTN)CpuMpData->CpuData[Index].StartupApSignal;
> + MpHandOff->Info[Index-HobBase].StartupProcedureAddress = (UINT64)(UINTN)&CpuMpData->CpuData[Index].ApFunction;
> }
> }
>
(3) The conversion looks good otherwise, but I dislike that
StartupSignalValue and WaitLoopExecutionMode get uselessly replicated
over all HOBs. Again, it *increases* technical debt. It's fine to ignore
existent debt (if you can), but adding to it is not right.
Can you file a new TianoCore BZ, for moving these fields to separate
dynamic PCDs, or to a new singleton GUID HOB (containing both fields)?
If you assign that BZ to yourself, and reference it in patches #4 and
#5, I'll be happy to R-b version 3 of the series. (Which is something
I'd like to do.)
Version 3 may be mergeable regardless, of course, if Ray accepts it.
Thanks,
Laszlo
-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#115770): https://edk2.groups.io/g/devel/message/115770
Mute This Topic: https://groups.io/mt/104472313/7686176
Group Owner: devel+owner@edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [rebecca@openfw.io]
-=-=-=-=-=-=-=-=-=-=-=-
prev parent reply other threads:[~2024-02-22 2:03 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-02-20 17:49 [edk2-devel] [PATCH v2 0/5] UefiCpuPkg/MpInitLib: Add support for multiple MP_HAND_OFF HOBs Gerd Hoffmann
2024-02-20 17:49 ` [edk2-devel] [PATCH v2 1/5] UefiCpuPkg/MpInitLib: Add support for multiple HOBs to GetMpHandOffHob Gerd Hoffmann
2024-02-21 3:35 ` Ni, Ray
2024-02-22 0:30 ` Laszlo Ersek
2024-02-20 17:49 ` [edk2-devel] [PATCH v2 2/5] UefiCpuPkg/MpInitLib: Add support for multiple HOBs to GetBspNumber() Gerd Hoffmann
2024-02-21 3:36 ` Ni, Ray
2024-02-21 3:36 ` Ni, Ray
2024-02-22 0:37 ` Laszlo Ersek
2024-02-22 0:38 ` Laszlo Ersek
2024-02-22 10:24 ` Ni, Ray
2024-02-20 17:49 ` [edk2-devel] [PATCH v2 3/5] UefiCpuPkg/MpInitLib: Add support for multiple HOBs to SwitchApContext() Gerd Hoffmann
2024-02-21 5:47 ` Ni, Ray
2024-02-22 0:44 ` Laszlo Ersek
2024-02-20 17:49 ` [edk2-devel] [PATCH v2 4/5] UefiCpuPkg/MpInitLib: Add support for multiple HOBs to MpInitLibInitialize Gerd Hoffmann
2024-02-21 3:37 ` Ni, Ray
2024-02-22 1:11 ` Laszlo Ersek
2024-02-22 12:28 ` Gerd Hoffmann
2024-02-23 0:23 ` Ni, Ray
2024-02-25 12:54 ` Laszlo Ersek
2024-02-20 17:49 ` [edk2-devel] [PATCH v2 5/5] UefiCpuPkg/MpInitLib: Add support for multiple HOBs to SaveCpuMpData() Gerd Hoffmann
2024-02-21 3:48 ` Ni, Ray
2024-02-21 10:24 ` Gerd Hoffmann
2024-02-22 1:34 ` Laszlo Ersek
2024-02-22 1:49 ` Laszlo Ersek
2024-02-22 2:17 ` Laszlo Ersek
2024-02-22 2:03 ` Laszlo Ersek [this message]
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=6ebafaae-25c6-e5c1-1c40-779cff5bfc24@redhat.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