From: "Wu, Hao A" <hao.a.wu@intel.com>
To: "Chiu, Ian" <Ian.chiu@intel.com>,
"devel@edk2.groups.io" <devel@edk2.groups.io>
Cc: "Chu, Maggie" <maggie.chu@intel.com>, "Ni, Ray" <ray.ni@intel.com>
Subject: Re: [PATCH] MdeModulePkg\UfsBlockIoPei: UFS MMIO address size support both 32/64 bit
Date: Tue, 26 Oct 2021 06:35:03 +0000 [thread overview]
Message-ID: <DM6PR11MB4025DF90EB0B783B1A2E3D2ECA849@DM6PR11MB4025.namprd11.prod.outlook.com> (raw)
In-Reply-To: <20211022091501.1991-1-ian.chiu@intel.com>
> -----Original Message-----
> From: Chiu, Ian <Ian.chiu@intel.com>
> Sent: Friday, October 22, 2021 5:15 PM
> To: devel@edk2.groups.io
> Cc: Chiu, Ian <Ian.chiu@intel.com>; Chiu, Ian <Ian.chiu@intel.com>; Chu,
> Maggie <maggie.chu@intel.com>; Ni, Ray <ray.ni@intel.com>; Wu, Hao A
> <hao.a.wu@intel.com>
> Subject: [PATCH] MdeModulePkg\UfsBlockIoPei: UFS MMIO address size
> support both 32/64 bit
>
> From: Ian Chiu <Ian.chiu@intel.com>
>
> https://bugzilla.tianocore.org/show_bug.cgi?id=3703
> MMIO base address size will overflow while finding two or more Host
> controller in the system. Correct it and support 32 and 64 bits address
> space.
Could you help to provide the information on what tests have been performed for this patch? Thanks.
Some additional inline comments below:
>
> Signed-off-by: Ian Chiu <ian.chiu@intel.com>
> Cc: Maggie Chu <maggie.chu@intel.com>
> Cc: Ray Ni <ray.ni@intel.com>
> Cc: Hao A Wu <hao.a.wu@intel.com>
> ---
> MdeModulePkg/Bus/Pci/UfsPciHcPei/UfsPciHcPei.c | 37
> ++++++++++++++++++++++++--
> 1 file changed, 35 insertions(+), 2 deletions(-)
>
> diff --git a/MdeModulePkg/Bus/Pci/UfsPciHcPei/UfsPciHcPei.c
> b/MdeModulePkg/Bus/Pci/UfsPciHcPei/UfsPciHcPei.c
> index 447a05b5b2..69a19c60a2 100644
> --- a/MdeModulePkg/Bus/Pci/UfsPciHcPei/UfsPciHcPei.c
> +++ b/MdeModulePkg/Bus/Pci/UfsPciHcPei/UfsPciHcPei.c
> @@ -76,6 +76,7 @@ InitializeUfsHcPeim (
> UINT16 Device;
>
> UINT16 Function;
>
> UINT32 Size;
>
> + UINT64 MmioSize;
>
> UINT8 SubClass;
>
> UINT8 BaseClass;
>
> UFS_HC_PEI_PRIVATE_DATA *Private;
>
> @@ -119,16 +120,48 @@ InitializeUfsHcPeim (
> PciAnd16 (PCI_LIB_ADDRESS (Bus, Device, Function,
> PCI_COMMAND_OFFSET), (UINT16)~(EFI_PCI_COMMAND_BUS_MASTER |
> EFI_PCI_COMMAND_MEMORY_SPACE));
>
> PciWrite32 (PCI_LIB_ADDRESS (Bus, Device, Function,
> PCI_BASE_ADDRESSREG_OFFSET), 0xFFFFFFFF);
>
> Size = PciRead32 (PCI_LIB_ADDRESS (Bus, Device, Function,
> PCI_BASE_ADDRESSREG_OFFSET));
>
> +
>
> + switch (Size & 0x07) {
>
> + case 0x0:
>
> + //
>
> + // Memory space: anywhere in 32 bit address space
>
> + //
>
> + MmioSize = (~(Size & 0xFFFFFFF0)) + 1;
>
> + break;
>
> + case 0x4:
>
> + //
>
> + // Memory space: anywhere in 64 bit address space
>
> + //
>
> + MmioSize = Size & 0xFFFFFFF0;
>
For 64-bit BAR, I think you also need to write 0xFFFFFFFF to the high 32-bit of the BAR and read the return value as well during the calculation of the request MMIO size.
> +
>
> + //
>
> + // Fix the length to support some spefic 64 bit BAR
Typo: spefic -> specific
>
> + //
>
> + Size |= ((UINT32)(-1) << HighBitSet32 (Size));
>
> +
>
> + //
>
> + // Calculate the size of 64bit bar
>
> + //
>
> + MmioSize |= LShiftU64 ((UINT64) Size, 32);
>
> + MmioSize = (~(MmioSize)) + 1;
With the above 64-bit BAR size change, I think you need to clean the high 32bits of this 64bit BAR to 0, since 32bit BAR address will be used in PEI phase.
>
> + break;
>
> + default:
>
> + //
>
> + // Unknown BAR type
>
> + //
>
> + ASSERT (FALSE);
>
> + continue;
>
> + };
>
> //
>
> // Assign resource to the Ufs Pci host controller's MMIO BAR.
>
> // Enable the Ufs Pci host controller by setting BME and MSE bits of
> PCI_CMD register.
>
> //
>
> - PciWrite32 (PCI_LIB_ADDRESS (Bus, Device, Function,
> PCI_BASE_ADDRESSREG_OFFSET), (UINT32)(PcdGet32
> (PcdUfsPciHostControllerMmioBase) + Size * Private->TotalUfsHcs));
>
> + PciWrite32 (PCI_LIB_ADDRESS (Bus, Device, Function,
> PCI_BASE_ADDRESSREG_OFFSET), (UINT32)(PcdGet32
> (PcdUfsPciHostControllerMmioBase) + MmioSize * Private->TotalUfsHcs));
I think the above line can be refined, the driver currently assumes that every UFS HC will have the same request MMIO size (using MmioSize * Private->TotalUfsHcs).
But I do not think this assumption will always be true, it is possible that UFS HCs will have different request MMIO size.
>
> PciOr16 (PCI_LIB_ADDRESS (Bus, Device, Function,
> PCI_COMMAND_OFFSET), (EFI_PCI_COMMAND_BUS_MASTER |
> EFI_PCI_COMMAND_MEMORY_SPACE));
>
> //
>
> // Record the allocated Mmio base address.
>
> //
>
> - Private->UfsHcPciAddr[Private->TotalUfsHcs] = PcdGet32
> (PcdUfsPciHostControllerMmioBase) + Size * Private->TotalUfsHcs;
>
> + Private->UfsHcPciAddr[Private->TotalUfsHcs] = PcdGet32
> (PcdUfsPciHostControllerMmioBase) + (UINTN) MmioSize * Private-
> >TotalUfsHcs;
Please update the above line accordingly (not assuming UFS HCs having the same request MMIO size).
Best Regards,
Hao Wu
>
> Private->TotalUfsHcs++;
>
> ASSERT (Private->TotalUfsHcs < MAX_UFS_HCS);
>
> }
>
> --
> 2.16.2.windows.1
prev parent reply other threads:[~2021-10-26 6:35 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-10-22 9:15 [PATCH] MdeModulePkg\UfsBlockIoPei: UFS MMIO address size support both 32/64 bit ian.chiu
2021-10-26 6:35 ` Wu, Hao A [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=DM6PR11MB4025DF90EB0B783B1A2E3D2ECA849@DM6PR11MB4025.namprd11.prod.outlook.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