public inbox for devel@edk2.groups.io
 help / color / mirror / Atom feed
From: "Wu, Hao A" <hao.a.wu@intel.com>
To: "He, Jiangang" <Jiangang.He@amd.com>,
	"Chang, Abner" <Abner.Chang@amd.com>,
	"devel@edk2.groups.io" <devel@edk2.groups.io>
Cc: "Lin, Kuei-Hung (Timothy)" <Kuei-Hung.Lin@amd.com>,
	"Ni, Ray" <ray.ni@intel.com>, "Zeng, Star" <star.zeng@intel.com>,
	"Sun, Zhikai" <zhikai.sun@intel.com>,
	"Kirkendall, Garrett" <Garrett.Kirkendall@amd.com>
Subject: Re: [PATCH V2] MdeModulePkg/XhciPei: Unlinked XhciPei memory block
Date: Tue, 20 Dec 2022 00:26:43 +0000	[thread overview]
Message-ID: <DM6PR11MB4025DCD4156BC784122E08D1CAEA9@DM6PR11MB4025.namprd11.prod.outlook.com> (raw)
In-Reply-To: <MN2PR12MB34219B5479B09A1117383CAFFDE59@MN2PR12MB3421.namprd12.prod.outlook.com>

Thanks.
You are right. I agree with you that aligning with XhciDxe is a better resolution.

Best Regards,
Hao Wu

> -----Original Message-----
> From: He, Jiangang <Jiangang.He@amd.com>
> Sent: Tuesday, December 20, 2022 6:54 AM
> To: Wu, Hao A <hao.a.wu@intel.com>; Chang, Abner
> <Abner.Chang@amd.com>; devel@edk2.groups.io
> Cc: Lin, Kuei-Hung (Timothy) <Kuei-Hung.Lin@amd.com>; Ni, Ray
> <ray.ni@intel.com>; Zeng, Star <star.zeng@intel.com>; Sun, Zhikai
> <zhikai.sun@intel.com>; Kirkendall, Garrett <Garrett.Kirkendall@amd.com>
> Subject: RE: [PATCH V2] MdeModulePkg/XhciPei: Unlinked XhciPei memory
> block
> 
> [AMD Official Use Only - General]
> 
> UsbHcFreeMemBlock()->IoMmuFreeBuffer()->mIoMmu->FreeBuffer(), which
> may end up calling PeiFreePages() depending on gEdkiiIoMmuPpiGuid
> implementation. Surely both will work since UsbHcFreeMemPool() can't be
> interrupted by any service call to use the memory just freed. Just for good
> coding practice reason, I pick the one aligning with XhciDxe.
> 
> Thanks,
> Jiangang
> -----Original Message-----
> From: Wu, Hao A <hao.a.wu@intel.com>
> Sent: Monday, December 19, 2022 12:40 AM
> To: He, Jiangang <Jiangang.He@amd.com>; Chang, Abner
> <Abner.Chang@amd.com>; devel@edk2.groups.io
> Cc: Lin, Kuei-Hung (Timothy) <Kuei-Hung.Lin@amd.com>; Ni, Ray
> <ray.ni@intel.com>; Zeng, Star <star.zeng@intel.com>; Sun, Zhikai
> <zhikai.sun@intel.com>; Kirkendall, Garrett <Garrett.Kirkendall@amd.com>
> Subject: RE: [PATCH V2] MdeModulePkg/XhciPei: Unlinked XhciPei memory
> block
> 
> [AMD Official Use Only - General]
> 
> Caution: This message originated from an External Source. Use proper
> caution when opening attachments, clicking links, or responding.
> 
> 
> Hello,
> 
> My take is that unlike in DXE, the UsbHcFreeMemBlock() implementation in
> PEI phase does not perform freeing the memory.
> 
> So I think both the solution:
> * Provided at
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fedk2
> .groups.io%2Fg%2Fdevel%2Ftopic%2F92833071%2392165&data=05%7C01%7
> CJiangang.He%40amd.com%7C528424aa303f4a2c9b1808dae18be036%7C3d
> d8961fe4884e608e11a82d994e183d%7C0%7C0%7C638070288126779782%7C
> Unknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBT
> iI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=uJMNyYXMU22
> UZNjrKtDetyD2BqrvkualuorPZHAV3wg%3D&reserved=0, which aligns with
> EhciPei
> * Mentioned at the end of your previous reply, which aligns with XhciDxe
> should work fine.
> 
> I will leave it to you for the final decision.
> 
> Best Regards,
> Hao Wu
> 
> > -----Original Message-----
> > From: He, Jiangang <Jiangang.He@amd.com>
> > Sent: Friday, December 16, 2022 12:48 AM
> > To: Chang, Abner <Abner.Chang@amd.com>; Wu, Hao A
> > <hao.a.wu@intel.com>; devel@edk2.groups.io
> > Cc: Lin, Kuei-Hung (Timothy) <Kuei-Hung.Lin@amd.com>; Ni, Ray
> > <ray.ni@intel.com>; Zeng, Star <star.zeng@intel.com>; Sun, Zhikai
> > <zhikai.sun@intel.com>; Kirkendall, Garrett
> > <Garrett.Kirkendall@amd.com>
> > Subject: RE: [PATCH V2] MdeModulePkg/XhciPei: Unlinked XhciPei memory
> > block
> >
> > [AMD Official Use Only - General]
> >
> > Yes, it is the same issue discussed in
> >
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fedk2
> > .groups.io%2Fg%2Fdevel%2Ftopic%2F92833071%2392165&data=05%7C01
> %7CJiang
> >
> ang.He%40amd.com%7C528424aa303f4a2c9b1808dae18be036%7C3dd8961f
> e4884e60
> >
> 8e11a82d994e183d%7C0%7C0%7C638070288126936018%7CUnknown%7CT
> WFpbGZsb3d8
> >
> eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3
> D%7C3
> >
> 000%7C%7C%7C&sdata=58j41QJxKbrQenhyZKYO4dxYj3Sat2kJejQGioZhtu4%3
> D&rese
> > rved=0
> >
> > MdeModulePkg\Bus\Pci\XhciPei\UsbHcMem.c
> >
> >   for (Block = Pool->Head->Next; Block != NULL; Block = Pool->Head->Next) {
> >     // UsbHcUnlinkMemBlock (Pool->Head, Block);
> >     UsbHcFreeMemBlock (Pool, Block);
> >   }
> > Block = Pool->Head->Next never change without calling
> > UsbHcUnlinkMemBlock (Pool->Head, Block), therefore dead loop.
> >
> > Our proposed fix came from dxe version of the equivalent file
> > MdeModulePkg\Bus\Pci\XhciDxe\UsbHcMem.c but swapped two routine
> call
> > order (Now I think it is incorrect as clarified below).
> >   for (Block = Pool->Head->Next; Block != NULL; Block = Pool->Head->Next) {
> >     UsbHcFreeMemBlock (Pool, Block);
> >     UsbHcUnlinkMemBlock (Pool->Head, Block);
> >   }
> >
> >
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fedk2
> .groups.io%2Fg%2Fdevel%2Ftopic%2F92833071%2392165&data=05%7C01%7
> CJiangang.He%40amd.com%7C528424aa303f4a2c9b1808dae18be036%7C3d
> d8961fe4884e608e11a82d994e183d%7C0%7C0%7C638070288126936018%7C
> Unknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBT
> iI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=58j41QJxKbrQen
> hyZKYO4dxYj3Sat2kJejQGioZhtu4%3D&reserved=0 proposed fix:
> >
> >   for (Block = Pool->Head->Next; Block != NULL; Block = Block ->Next) {
> >     // UsbHcUnlinkMemBlock (Pool->Head, Block);
> >     UsbHcFreeMemBlock (Pool, Block);
> >   }
> >
> > I think it again, both proposals have problem of reading memory
> > content in the buffer that has just been freed.
> >
> >   for (Block = Pool->Head->Next; Block != NULL; Block = Pool->Head->Next) {
> >      UsbHcUnlinkMemBlock (Pool->Head, Block);
> >     UsbHcFreeMemBlock (Pool, Block);
> >   }
> > is right solution and matches dxe version of UsbHcMem.c.
> >
> > Thanks,
> > Jiangang
> >
> > -----Original Message-----
> > From: Chang, Abner <Abner.Chang@amd.com>
> > Sent: Wednesday, December 14, 2022 8:12 PM
> > To: Wu, Hao A <hao.a.wu@intel.com>; devel@edk2.groups.io
> > Cc: Lin, Kuei-Hung (Timothy) <Kuei-Hung.Lin@amd.com>; Ni, Ray
> > <ray.ni@intel.com>; Zeng, Star <star.zeng@intel.com>; Sun, Zhikai
> > <zhikai.sun@intel.com>; Kirkendall, Garrett
> > <Garrett.Kirkendall@amd.com>; He, Jiangang <Jiangang.He@amd.com>
> > Subject: RE: [PATCH V2] MdeModulePkg/XhciPei: Unlinked XhciPei memory
> > block
> >
> > [AMD Official Use Only - General]
> >
> > Hi Jiangang,
> > Could you please provide the context of this patch?
> >
> > Thanks
> > Abner
> >
> > > -----Original Message-----
> > > From: Wu, Hao A <hao.a.wu@intel.com>
> > > Sent: Monday, December 12, 2022 11:27 AM
> > > To: Chang, Abner <Abner.Chang@amd.com>; devel@edk2.groups.io
> > > Cc: Lin, Kuei-Hung (Timothy) <Kuei-Hung.Lin@amd.com>; Ni, Ray
> > > <ray.ni@intel.com>; Zeng, Star <star.zeng@intel.com>; Sun, Zhikai
> > > <zhikai.sun@intel.com>; Kirkendall, Garrett
> > > <Garrett.Kirkendall@amd.com>
> > > Subject: RE: [PATCH V2] MdeModulePkg/XhciPei: Unlinked XhciPei
> > > memory block
> > >
> > > Caution: This message originated from an External Source. Use proper
> > > caution when opening attachments, clicking links, or responding.
> > >
> > >
> > > Sorry for a question, may I know what issue was met that leads to
> > > the proposed patch?
> > > Could you help to check if it is related with the topic discussed in
> > >
> >
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fedk2
> > > .gr
> > >
> > oups.io%2Fg%2Fdevel%2Ftopic%2F92833071%2392165&amp;data=05%7C0
> > 1%7
> > >
> >
> Cabner.chang%40amd.com%7Ccac49c2820e741b8c48a08dadbf0cc24%7C3d
> > d8
> > >
> > 961fe4884e608e11a82d994e183d%7C0%7C0%7C638064124512265992%7C
> > Unk
> > >
> >
> nown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6I
> > k1h
> > >
> >
> aWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&amp;sdata=z1Q7NRxN4GMA
> > %2
> > > FBxYd2D7Gnkc3aTD23mRnwNF3H5wE0k%3D&amp;reserved=0? Thanks
> in
> > advance.
> > >
> > > Best Regards,
> > > Hao Wu
> > >
> > > > -----Original Message-----
> > > > From: abner.chang@amd.com <abner.chang@amd.com>
> > > > Sent: Saturday, December 10, 2022 11:13 PM
> > > > To: devel@edk2.groups.io
> > > > Cc: Kuei-Hung.Lin@amd.com; Wu, Hao A <hao.a.wu@intel.com>; Ni,
> Ray
> > > > <ray.ni@intel.com>; Garrett Kirkendall
> > > > <garrett.kirkendall@amd.com>; Abner Chang <abner.chang@amd.com>
> > > > Subject: [PATCH V2] MdeModulePkg/XhciPei: Unlinked XhciPei memory
> > > > block
> > > >
> > > > From: Abner Chang <abner.chang@amd.com>
> > > >
> > > > In V2: Add AMD copyright.
> > > >
> > > > Unlink the XhciPei memory block when it has been freed.
> > > >
> > > > Signed-off-by: Kuei-Hung Lin <Kuei-Hung.Lin@amd.com>
> > > > Cc: Hao A Wu <hao.a.wu@intel.com>
> > > > Cc: Ray Ni <ray.ni@intel.com>
> > > > Cc: Garrett Kirkendall <garrett.kirkendall@amd.com>
> > > > Cc: Abner Chang <abner.chang@amd.com>
> > > > ---
> > > >  MdeModulePkg/Bus/Pci/XhciPei/UsbHcMem.c | 29
> > > > ++++++++++++++++++++++++-
> > > >  1 file changed, 28 insertions(+), 1 deletion(-)
> > > >
> > > > diff --git a/MdeModulePkg/Bus/Pci/XhciPei/UsbHcMem.c
> > > > b/MdeModulePkg/Bus/Pci/XhciPei/UsbHcMem.c
> > > > index c64b38fcfc8..39ba31b0913 100644
> > > > --- a/MdeModulePkg/Bus/Pci/XhciPei/UsbHcMem.c
> > > > +++ b/MdeModulePkg/Bus/Pci/XhciPei/UsbHcMem.c
> > > > @@ -3,6 +3,7 @@ PEIM to produce gPeiUsb2HostControllerPpiGuid
> > based
> > > > on gPeiUsbControllerPpiGuid  which is used to enable recovery
> > > > function from USB Drivers.
> > > >
> > > >  Copyright (c) 2014 - 2016, Intel Corporation. All rights
> > > > reserved.<BR>
> > > > +Copyright (C) 2022 Advanced Micro Devices, Inc. All rights
> > > > +reserved.<BR>
> > > >
> > > >  SPDX-License-Identifier: BSD-2-Clause-Patent
> > > >
> > > > @@ -365,6 +366,32 @@ UsbHcInitMemPool (
> > > >    return Pool;
> > > >  }
> > > >
> > > > +/**
> > > > +  Unlink the memory block from the pool's list.
> > > > +
> > > > +  @param  Head           The block list head of the memory's pool.
> > > > +  @param  BlockToUnlink  The memory block to unlink.
> > > > +
> > > > +**/
> > > > +VOID
> > > > +UsbHcUnlinkMemBlock (
> > > > +  IN USBHC_MEM_BLOCK  *Head,
> > > > +  IN USBHC_MEM_BLOCK  *BlockToUnlink
> > > > +  )
> > > > +{
> > > > +  USBHC_MEM_BLOCK  *Block;
> > > > +
> > > > +  ASSERT ((Head != NULL) && (BlockToUnlink != NULL));
> > > > +
> > > > +  for (Block = Head; Block != NULL; Block = Block->Next) {
> > > > +    if (Block->Next == BlockToUnlink) {
> > > > +      Block->Next         = BlockToUnlink->Next;
> > > > +      BlockToUnlink->Next = NULL;
> > > > +      break;
> > > > +    }
> > > > +  }
> > > > +}
> > > > +
> > > >  /**
> > > >    Release the memory management pool.
> > > >
> > > > @@ -386,8 +413,8 @@ UsbHcFreeMemPool (
> > > >    // first block.
> > > >    //
> > > >    for (Block = Pool->Head->Next; Block != NULL; Block =
> > > > Pool->Head->Next) {
> > > > -    // UsbHcUnlinkMemBlock (Pool->Head, Block);
> > > >      UsbHcFreeMemBlock (Pool, Block);
> > > > +    UsbHcUnlinkMemBlock (Pool->Head, Block);
> > > >    }
> > > >
> > > >    UsbHcFreeMemBlock (Pool, Pool->Head);
> > > > --
> > > > 2.37.1.windows.1

  reply	other threads:[~2022-12-20  0:26 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-12-10 15:12 [PATCH V2] MdeModulePkg/XhciPei: Unlinked XhciPei memory block Chang, Abner
2022-12-12  3:27 ` Wu, Hao A
2022-12-15  2:11   ` Chang, Abner
2022-12-15 16:47     ` He, Jiangang
2022-12-19  6:40       ` Wu, Hao A
2022-12-19 22:53         ` He, Jiangang
2022-12-20  0:26           ` Wu, Hao A [this message]
2022-12-20  0:27             ` Chang, Abner

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=DM6PR11MB4025DCD4156BC784122E08D1CAEA9@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