public inbox for devel@edk2.groups.io
 help / color / mirror / Atom feed
From: Jordan Justen <jordan.l.justen@intel.com>
To: "Ni, Ray" <ray.ni@Intel.com>, Andrew Fish <afish@apple.com>,
	Laszlo Ersek <lersek@redhat.com>, Liu Yu <pedroa.liu@outlook.com>
Cc: "edk2-devel@lists.01.org" <edk2-devel@lists.01.org>
Subject: Re: EmulatorPkg Unix Host Segmentation fault.
Date: Sat, 16 Feb 2019 01:11:10 -0800	[thread overview]
Message-ID: <155030827034.1052.16419444135424097946@jljusten-skl> (raw)
In-Reply-To: <f16bb263-1c81-7957-4789-b3f3113266f8@Intel.com>

On 2019-02-16 00:05:27, Ni, Ray wrote:
> On 2/16/2019 3:43 PM, Ni, Ray wrote:
> > (Sent third times to make sure Andrew and Laszlo are in the TO list.)
> > 
> > I also met this issue.
> > I found three solutions:
> > 1. Forcing PeiMain CC flag to "-O0" works.
> > 2. Changing EmulatorPkg/Sec to not produce TemporaryRamSupportPpi also 
> > works.
> > 3. Implement the temporary migration routine as below in EmulatorPkg/Sec 
> > module.
> > 
> > EFI_STATUS
> > EFIAPI
> > SecTemporaryRamSupport (
> >    IN CONST EFI_PEI_SERVICES   **PeiServices,
> >    IN EFI_PHYSICAL_ADDRESS     TemporaryMemoryBase,
> >    IN EFI_PHYSICAL_ADDRESS     PermanentMemoryBase,
> >    IN UINTN                    CopySize
> >    )
> > {
> >    VOID                             *OldHeap;
> >    VOID                             *NewHeap;
> >    VOID                             *OldStack;
> >    VOID                             *NewStack;
> >    UINTN                            StackMigrateOffset;
> >    BASE_LIBRARY_JUMP_BUFFER         JumpBuffer;
> > 
> >    DEBUG ((EFI_D_INFO,
> >      "TemporaryRamMigration(0x%Lx, 0x%Lx, 0x%Lx)\n",
> >      TemporaryMemoryBase,
> >      PermanentMemoryBase,
> >      (UINT64)CopySize
> >      ));
> > 
> >    //
> >    // Assume Host prepare the stack and heap in the temprary ram that stack
> >    // is below heap (stack is in smaller address).
> >    // Stack/heap migration depends on the stack/heap location information
> >    // in the temporary ram.
> >    //
> >    OldStack = (VOID*)(UINTN)TemporaryMemoryBase;
> >    NewStack = (VOID*)((UINTN)PermanentMemoryBase);
> > 
> >    OldHeap = (VOID*)((UINTN)TemporaryMemoryBase + (CopySize >> 1));
> >    NewHeap = (VOID*)((UINTN)PermanentMemoryBase + (CopySize >> 1));
> > 
> >    StackMigrateOffset = (UINTN)NewStack - (UINTN)OldStack;
> > 
> >    //
> >    // Migrate Heap and Stack
> >    //
> >    CopyMem (NewHeap, OldHeap, CopySize >> 1);
> >    CopyMem (NewStack, OldStack, CopySize >> 1);
> > 
> >    //
> >    // Use SetJump()/LongJump() to switch to a new stack.
> >    //
> >    if (SetJump (&JumpBuffer) == 0) {
> > #if defined (MDE_CPU_IA32)
> >      JumpBuffer.Esp = JumpBuffer.Esp + StackMigrateOffset;
> >      JumpBuffer.Ebp = JumpBuffer.Ebp + StackMigrateOffset;
> > #endif
> > #if defined (MDE_CPU_X64)
> >      JumpBuffer.Rsp = JumpBuffer.Rsp + StackMigrateOffset;
> >      JumpBuffer.Rbp = JumpBuffer.Rbp + StackMigrateOffset;
> > #endif
> >      LongJump (&JumpBuffer, (UINTN)-1);
> >    }
> > 
> >    ZeroMem ((VOID *)(UINTN) TemporaryMemoryBase, CopySize);
> > 
> >    return EFI_SUCCESS;
> > }
> > 
> > Andrew,
> > I'd like to know why you chose to produce the migration PPI from
> > EmulatorPkg/Sec module.
> > Based on PI spec and current PeiCore implementation, PeiCore can do the 
> > migration when PPI is absent.
> > 
> Study the PeiCore migration logic a bit more, I found since PeiCore 
> knows the exact size of new stack in permanent memory, it migrates the 
> old stack to the top of new stack.
> But the migration logic in above C code (since it doesn't know the size 
> of new stack, CopySize is the size of temporary memory) may copy the old 
> stack to the middle in new stack.

We had a fair amount of discussion about this in November.

Search for "EmulatorPkg Unix" on
https://lists.01.org/pipermail/edk2-devel/2018-November/thread.html.

I have a branch related to that discussion:

https://github.com/jljusten/edk2/tree/temp-ram-support2

There are essentially 2 parts to the branch. First is a fix for the
PEI Core TemporaryRamMigration implementation.

After the "MdeModulePkg/Core/Pei: Use assembly for X64
TemporaryRamMigration" patch on that branch, I was working on a new
TemporaryRamMigration2 PPI that should fix the issues with
TemporaryRamMigration. Of course, that would have to become part of
the standard before EDK-II could add it.

-Jordan


  reply	other threads:[~2019-02-16  9:11 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-11-18  4:51 EmulatorPkg Unix Host Segmentation fault Liu Yu
2018-11-18  9:27 ` Jordan Justen
2018-11-18 12:07   ` Liu Yu
2018-11-18 22:37     ` Andrew Fish
2018-11-19  1:13       ` Jordan Justen
2018-11-19 19:16         ` Jordan Justen
2018-11-19 21:22           ` Andrew Fish
2018-11-19 22:12             ` Laszlo Ersek
2018-11-19 23:39               ` Andrew Fish
2018-11-19 22:29             ` Jordan Justen
2018-11-20  0:54               ` Andrew Fish
2018-11-20  8:57               ` Laszlo Ersek
2019-02-16  7:29 ` Ni, Ray
2019-02-16  7:40 ` Ni, Ray
2019-02-16 20:23   ` Andrew Fish
2019-02-16  7:43 ` Ni, Ray
2019-02-16  8:05   ` Ni, Ray
2019-02-16  9:11     ` Jordan Justen [this message]
2019-02-18  2:25       ` Ni, Ray
2019-02-18  2:45         ` Jordan Justen

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=155030827034.1052.16419444135424097946@jljusten-skl \
    --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