From: "Ard Biesheuvel" <ard.biesheuvel@arm.com>
To: Leif Lindholm <leif@nuviainc.com>
Cc: devel@edk2.groups.io, liming.gao@intel.com, lersek@redhat.com,
Hao A Wu <hao.a.wu@intel.com>, Ray Ni <ray.ni@intel.com>
Subject: Re: [PATCH] MdeModulePkg/NonDiscoverablePciDeviceDxe: connect handles on registration
Date: Thu, 21 May 2020 19:29:33 +0200 [thread overview]
Message-ID: <87f6f6a9-7f99-d779-09e5-7f36fead3731@arm.com> (raw)
In-Reply-To: <20200521111651.GR1923@vanye>
On 5/21/20 1:16 PM, Leif Lindholm wrote:
> On Thu, May 21, 2020 at 13:10:28 +0200, Ard Biesheuvel wrote:
>> The way EDK2 invokes the UEFI driver model assumes that PCI I/O
>> protocol instances exist for all PCI I/O controllers in the system.
>>
>> For instance, UefiBootManagerLib connects the short-form USB device
>> path of the console input by looking for PCI I/O controllers that
>> have the 'USB host controller' class code, and passing each one to
>> ConnectController(), using the device path as the 'RemainingDevicePath'
>> argument.
>>
>> For true PCI I/O protocol instances produced by the PCI root bridge
>> driver, this works fine, since it always enumerates the PCIe hierarchy
>> exhaustively. However, for platform devices that are wired to PCI class
>> drivers using the non-discoverable PCIe driver, this breaks down, due
>> to the fact that the PCI I/O protocol instance does not exist unless the
>> non-discoverable device protocol handle is connected first.
>>
>> So let's connect these handles non-recursively as soon as they appear.
>>
>> Cc: Hao A Wu <hao.a.wu@intel.com>
>> Cc: Ray Ni <ray.ni@intel.com>
>> Signed-off-by: Ard Biesheuvel <ard.biesheuvel@arm.com>
>
> Acked-by: Leif Lindholm <leif@nuviainc.com>
>
Thanks,
I realized that this version is broken, though: passing gImageHandle as
*DriverImageHandle is not correct, it should be an array { gImageHandle,
NULL } instead.
>> ---
>> MdeModulePkg/Bus/Pci/NonDiscoverablePciDeviceDxe/NonDiscoverablePciDeviceDxe.c | 43 ++++++++++++++++++++
>> 1 file changed, 43 insertions(+)
>>
>> diff --git a/MdeModulePkg/Bus/Pci/NonDiscoverablePciDeviceDxe/NonDiscoverablePciDeviceDxe.c b/MdeModulePkg/Bus/Pci/NonDiscoverablePciDeviceDxe/NonDiscoverablePciDeviceDxe.c
>> index 5c93e2a7663c..a14c06e7f4e1 100644
>> --- a/MdeModulePkg/Bus/Pci/NonDiscoverablePciDeviceDxe/NonDiscoverablePciDeviceDxe.c
>> +++ b/MdeModulePkg/Bus/Pci/NonDiscoverablePciDeviceDxe/NonDiscoverablePciDeviceDxe.c
>> @@ -15,6 +15,8 @@
>> STATIC UINTN mUniqueIdCounter = 0;
>> EFI_CPU_ARCH_PROTOCOL *mCpu;
>>
>> +STATIC VOID *mProtocolNotifyRegistration;
>> +
>> //
>> // We only support the following device types
>> //
>> @@ -250,6 +252,43 @@ STATIC EFI_DRIVER_BINDING_PROTOCOL gDriverBinding = {
>> NULL
>> };
>>
>> +STATIC
>> +VOID
>> +EFIAPI
>> +NonDiscoverablePciDeviceProtocolNotify (
>> + IN EFI_EVENT Event,
>> + IN VOID *Context
>> + )
>> +{
>> + EFI_STATUS Status;
>> + EFI_HANDLE *Handles;
>> + UINTN HandleCount;
>> + UINTN Index;
>> +
>> + Status = gBS->LocateHandleBuffer (ByRegisterNotify, NULL,
>> + mProtocolNotifyRegistration, &HandleCount, &Handles);
>> + if (EFI_ERROR (Status)) {
>> + if (Status != EFI_NOT_FOUND) {
>> + DEBUG ((DEBUG_WARN, "%a: LocateHandleBuffer() failed - %r\n",
>> + __FUNCTION__, Status));
>> + }
>> + return;
>> + }
>> +
>> + for (Index = 0; Index < HandleCount; Index++) {
>> + //
>> + // Connect each newly registered gEdkiiNonDiscoverableDeviceProtocolGuid
>> + // instance non-recursively to this driver specifically. This ensures that
>> + // PCI I/O instances exist for each, regardless of whether ConnectAll() is
>> + // used at any point.
>> + //
>> + Status = gBS->ConnectController (Handles[Index], gImageHandle, NULL, FALSE);
>> + DEBUG ((DEBUG_VERBOSE, "%a: ConnectController () returned %r\n",
>> + __FUNCTION__, Status));
>> + }
>> + gBS->FreePool (Handles);
>> +}
>> +
>> /**
>> Entry point of this driver.
>>
>> @@ -272,6 +311,10 @@ NonDiscoverablePciDeviceDxeEntryPoint (
>> Status = gBS->LocateProtocol (&gEfiCpuArchProtocolGuid, NULL, (VOID **)&mCpu);
>> ASSERT_EFI_ERROR(Status);
>>
>> + EfiCreateProtocolNotifyEvent (&gEdkiiNonDiscoverableDeviceProtocolGuid,
>> + TPL_CALLBACK, NonDiscoverablePciDeviceProtocolNotify, NULL,
>> + &mProtocolNotifyRegistration);
>> +
>> return EfiLibInstallDriverBindingComponentName2 (
>> ImageHandle,
>> SystemTable,
>> --
>> 2.17.1
>>
next prev parent reply other threads:[~2020-05-21 17:29 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-05-21 11:10 [PATCH] MdeModulePkg/NonDiscoverablePciDeviceDxe: connect handles on registration Ard Biesheuvel
2020-05-21 11:16 ` Leif Lindholm
2020-05-21 17:29 ` Ard Biesheuvel [this message]
2020-05-21 21:12 ` Laszlo Ersek
2020-05-21 21:58 ` [edk2-devel] " Ard Biesheuvel
2020-05-22 16:36 ` Laszlo Ersek
2020-05-22 16:46 ` Laszlo Ersek
2020-05-22 16:48 ` Laszlo Ersek
2020-05-22 17:04 ` 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-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=87f6f6a9-7f99-d779-09e5-7f36fead3731@arm.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