From: "Dong, Eric" <eric.dong@intel.com>
To: Laszlo Ersek <lersek@redhat.com>,
"devel@edk2.groups.io" <devel@edk2.groups.io>
Cc: "Ni, Ray" <ray.ni@intel.com>
Subject: Re: [Patch] UefiCpuPkg/CpuDxe: clean up PAGE_TABLE_LIB_PAGING_CONTEXT usage.
Date: Thu, 12 Sep 2019 02:54:33 +0000 [thread overview]
Message-ID: <ED077930C258884BBCB450DB737E662259EE48A5@shsmsx102.ccr.corp.intel.com> (raw)
In-Reply-To: <e7d70771-66da-18b1-49f4-f39ba6790798@redhat.com>
Hi Laszlo,
> -----Original Message-----
> From: Laszlo Ersek [mailto:lersek@redhat.com]
> Sent: Wednesday, September 11, 2019 11:25 PM
> To: Dong, Eric <eric.dong@intel.com>; devel@edk2.groups.io
> Cc: Ni, Ray <ray.ni@intel.com>
> Subject: Re: [Patch] UefiCpuPkg/CpuDxe: clean up
> PAGE_TABLE_LIB_PAGING_CONTEXT usage.
>
> On 09/11/19 03:45, Eric Dong wrote:
> > REF: https://bugzilla.tianocore.org/show_bug.cgi?id=1039
> >
> > Current implementation not checks system mode before using
> > PAGE_TABLE_LIB_PAGING_CONTEXT.ContextData.X64 or
> > PAGE_TABLE_LIB_PAGING_CONTEXT.ContextData.Ia32. This patch check
> the
> > mode before using the correct one.
> >
> > Cc: Ray Ni <ray.ni@intel.com>
> > Cc: Laszlo Ersek <lersek@redhat.com>
> > Signed-off-by: Eric Dong <eric.dong@intel.com>
> > ---
> > UefiCpuPkg/CpuDxe/CpuPageTable.c | 52
> > +++++++++++++++++++++++---------
> > 1 file changed, 37 insertions(+), 15 deletions(-)
> >
> > diff --git a/UefiCpuPkg/CpuDxe/CpuPageTable.c
> > b/UefiCpuPkg/CpuDxe/CpuPageTable.c
> > index ec5cd424fc..308f93b1cd 100644
> > --- a/UefiCpuPkg/CpuDxe/CpuPageTable.c
> > +++ b/UefiCpuPkg/CpuDxe/CpuPageTable.c
> > @@ -155,6 +155,8 @@ GetCurrentPagingContext (
> > MSR_IA32_EFER_REGISTER MsrEfer;
> > IA32_CR4 Cr4;
> > IA32_CR0 Cr0;
> > + UINT32 *Attributes;
> > + UINTN *PageTableBase;
> >
> > //
> > // Don't retrieve current paging context from processor if in SMM mode.
> > @@ -163,29 +165,33 @@ GetCurrentPagingContext (
> > ZeroMem (&mPagingContext, sizeof(mPagingContext));
> > if (sizeof(UINTN) == sizeof(UINT64)) {
> > mPagingContext.MachineType = IMAGE_FILE_MACHINE_X64;
> > + Attributes = &mPagingContext.ContextData.X64.Attributes;
> > + PageTableBase = &mPagingContext.ContextData.X64.PageTableBase;
> > } else {
> > mPagingContext.MachineType = IMAGE_FILE_MACHINE_I386;
> > + Attributes = &mPagingContext.ContextData.Ia32.Attributes;
> > + PageTableBase = (UINTN
> > + *)&mPagingContext.ContextData.Ia32.PageTableBase;
>
> (1) This is quite unfortunate. I don't like the cast.
>
> I understand why it is used here -- when we build for X64, then UINTN is
> UINT64, and so PageTableBase points to a UINT64. But
> "mPagingContext.ContextData.Ia32.PageTableBase" is a UINT32, so the
> compiler will complain about the incompatible pointer assignment.
> Therefore you use an explicit cast -- which would be wrong, if the code ever ran,
> but then again, that code will never run.
>
> I don't like this trick.
>
> Furthermore, the same issue will exist on the *other* branch, if you build the
> present code for IA32 -- you didn't use a cast there. When building for IA32,
> PageTableBase points to a UINT32. But, on the first branch,
> "mPagingContext.ContextData.X64.PageTableBase" is a UINT64.
> Therefore, the compiler will warn the same (although the code will never run).
>
> Did you build the patch for IA32?
Yes, I have got a report said IA32 and GCC5 build failure. I have prepared V2 change to fix this issue. I don't want to send too much version of code change, so I wait for some feedback from this mail list before I send v2 change.
>
>
> I think there are two ways to fix this.
>
> (1a) The ugly (but correct) fix is to use "#ifdef MDE_CPU_IA32" and "#ifdef
> MDE_CPU_X64" directives, and *no* casts. In other words, we should never
> compile the code that we won't run.
>
> (1b) The nice (and also correct) fix is to introduce a helper function:
>
> VOID
> GetPagingDetails (
> IN PAGE_TABLE_LIB_PAGING_CONTEXT_DATA *PagingContextData,
> OUT UINTN **PageTableBase OPTIONAL,
> OUT UINT32 **Attributes OPTIONAL
> );
>
> There would be two implementations:
>
> - IA32 subdirectory:
>
> if (PageTableBase != NULL) {
> *PageTableBase = &PagingContextData->Ia32.PageTableBase;
> }
> if (Attributes != NULL) {
> *Attributes = &PagingContextData->Ia32.Attributes;
> }
>
> - X64 subdirectory:
>
> if (PageTableBase != NULL) {
> *PageTableBase = &PagingContextData->X64.PageTableBase;
> }
> if (Attributes != NULL) {
> *Attributes = &PagingContextData->X64.Attributes;
> }
>
> And here we'd do
>
> GetPagingDetails (
> &mPagingContext.ContextData,
> &PageTableBase,
> &Attributes
> );
>
> and the rest of the GetCurrentPagingContext() updates would be the same.
>
I will use this option to update my code.
>
> > }
> >
> > Cr0.UintN = AsmReadCr0 ();
> > Cr4.UintN = AsmReadCr4 ();
> >
> > if (Cr0.Bits.PG != 0) {
> > - mPagingContext.ContextData.X64.PageTableBase = (AsmReadCr3 () &
> PAGING_4K_ADDRESS_MASK_64);
> > + *PageTableBase = (AsmReadCr3 () & PAGING_4K_ADDRESS_MASK_64);
> > } else {
> > - mPagingContext.ContextData.X64.PageTableBase = 0;
> > + *PageTableBase = 0;
> > }
> > if (Cr0.Bits.WP != 0) {
> > - mPagingContext.ContextData.Ia32.Attributes |=
> PAGE_TABLE_LIB_PAGING_CONTEXT_IA32_X64_ATTRIBUTES_WP_ENABLE;
> > + *Attributes |=
> > +
> PAGE_TABLE_LIB_PAGING_CONTEXT_IA32_X64_ATTRIBUTES_WP_ENABLE;
> > }
> > if (Cr4.Bits.PSE != 0) {
> > - mPagingContext.ContextData.Ia32.Attributes |=
> PAGE_TABLE_LIB_PAGING_CONTEXT_IA32_X64_ATTRIBUTES_PSE;
> > + *Attributes |=
> > + PAGE_TABLE_LIB_PAGING_CONTEXT_IA32_X64_ATTRIBUTES_PSE;
> > }
> > if (Cr4.Bits.PAE != 0) {
> > - mPagingContext.ContextData.Ia32.Attributes |=
> PAGE_TABLE_LIB_PAGING_CONTEXT_IA32_X64_ATTRIBUTES_PAE;
> > + *Attributes |=
> > + PAGE_TABLE_LIB_PAGING_CONTEXT_IA32_X64_ATTRIBUTES_PAE;
> > }
> > if (Cr4.Bits.LA57 != 0) {
> > - mPagingContext.ContextData.Ia32.Attributes |=
> PAGE_TABLE_LIB_PAGING_CONTEXT_IA32_X64_ATTRIBUTES_5_LEVEL;
> > + *Attributes |=
> > + PAGE_TABLE_LIB_PAGING_CONTEXT_IA32_X64_ATTRIBUTES_5_LEVEL;
> > }
> >
> > AsmCpuid (CPUID_EXTENDED_FUNCTION, &RegEax, NULL, NULL, NULL);
> @@
> > -197,12 +203,12 @@ GetCurrentPagingContext (
> > MsrEfer.Uint64 = AsmReadMsr64(MSR_CORE_IA32_EFER);
> > if (MsrEfer.Bits.NXE != 0) {
> > // XD activated
> > - mPagingContext.ContextData.Ia32.Attributes |=
> PAGE_TABLE_LIB_PAGING_CONTEXT_IA32_X64_ATTRIBUTES_XD_ACTIVATE
> D;
> > + *Attributes |=
> > +
> PAGE_TABLE_LIB_PAGING_CONTEXT_IA32_X64_ATTRIBUTES_XD_ACTIVATE
> D;
> > }
> > }
> >
> > if (RegEdx.Bits.Page1GB != 0) {
> > - mPagingContext.ContextData.Ia32.Attributes |=
> PAGE_TABLE_LIB_PAGING_CONTEXT_IA32_X64_ATTRIBUTES_PAGE_1G_SUP
> PORT;
> > + *Attributes |=
> > +
> PAGE_TABLE_LIB_PAGING_CONTEXT_IA32_X64_ATTRIBUTES_PAGE_1G_SUP
> PORT;
> > }
> > }
> > }
> > @@ -395,6 +401,7 @@ ConvertPageEntryAttribute ( {
> > UINT64 CurrentPageEntry;
> > UINT64 NewPageEntry;
> > + UINT32 *PageAttributes;
> >
> > CurrentPageEntry = *PageEntry;
> > NewPageEntry = CurrentPageEntry;
> > @@ -438,7 +445,13 @@ ConvertPageEntryAttribute (
> > break;
> > }
> > }
> > - if ((PagingContext->ContextData.Ia32.Attributes &
> >
> PAGE_TABLE_LIB_PAGING_CONTEXT_IA32_X64_ATTRIBUTES_XD_ACTIVATE
> D) != 0)
> > {
> > +
> > + if (PagingContext->MachineType == IMAGE_FILE_MACHINE_X64) {
> > + PageAttributes = &PagingContext->ContextData.X64.Attributes;
> > + } else {
> > + PageAttributes = &PagingContext->ContextData.Ia32.Attributes;
> > + }
>
> (2) So here we'd write
>
> GetPagingDetails (
> &PagingContext->ContextData,
> NULL,
> &PageAttributes
> );
>
> > + if ((*PageAttributes &
> > +
> PAGE_TABLE_LIB_PAGING_CONTEXT_IA32_X64_ATTRIBUTES_XD_ACTIVATE
> D) !=
> > + 0) {
> > if ((Attributes & EFI_MEMORY_XP) != 0) {
> > switch (PageAction) {
> > case PageActionAssign:
> > @@ -1338,15 +1351,24 @@ InitializePageTableLib (
> > )
> > {
> > PAGE_TABLE_LIB_PAGING_CONTEXT CurrentPagingContext;
> > + UINT32 *Attributes;
> > + UINTN *PageTableBase;
> >
> > GetCurrentPagingContext (&CurrentPagingContext);
> >
> > + if (CurrentPagingContext.MachineType == IMAGE_FILE_MACHINE_X64) {
> > + Attributes = &CurrentPagingContext.ContextData.X64.Attributes;
> > + PageTableBase =
> > + &CurrentPagingContext.ContextData.X64.PageTableBase;
> > + } else {
> > + Attributes = &CurrentPagingContext.ContextData.Ia32.Attributes;
> > + PageTableBase = (UINTN
> > + *)&CurrentPagingContext.ContextData.Ia32.PageTableBase;
> > + }
> > +
>
> (3) This would become
>
> GetPagingDetails (
> &CurrentPagingContext.ContextData,
> &PageTableBase,
> &Attributes
> );
>
> > //
> > // Reserve memory of page tables for future uses, if paging is enabled.
> > //
> > - if (CurrentPagingContext.ContextData.X64.PageTableBase != 0 &&
> > - (CurrentPagingContext.ContextData.Ia32.Attributes &
> > - PAGE_TABLE_LIB_PAGING_CONTEXT_IA32_X64_ATTRIBUTES_PAE) != 0)
> {
> > + if ((*PageTableBase != 0) &&
> > + (*Attributes &
> > + PAGE_TABLE_LIB_PAGING_CONTEXT_IA32_X64_ATTRIBUTES_PAE) != 0) {
> > DisableReadOnlyPageWriteProtect ();
> > InitializePageTablePool (1);
> > EnableReadOnlyPageWriteProtect (); @@ -1361,10 +1383,10 @@
> > InitializePageTableLib (
> > ASSERT (mLastPFEntryPointer != NULL);
> > }
> >
> > - DEBUG ((DEBUG_INFO, "CurrentPagingContext:\n",
> > CurrentPagingContext.MachineType));
> > + DEBUG ((DEBUG_INFO, "CurrentPagingContext:\n"));
> > DEBUG ((DEBUG_INFO, " MachineType - 0x%x\n",
> CurrentPagingContext.MachineType));
> > - DEBUG ((DEBUG_INFO, " PageTableBase - 0x%x\n",
> CurrentPagingContext.ContextData.X64.PageTableBase));
> > - DEBUG ((DEBUG_INFO, " Attributes - 0x%x\n",
> CurrentPagingContext.ContextData.X64.Attributes));
> > + DEBUG ((DEBUG_INFO, " PageTableBase - 0x%x\n", *PageTableBase));
>
> (4) UINTN values should be cast to UINT64 explicitly, and printed with
> "%Lx":
>
> DEBUG ((
> DEBUG_INFO,
> " PageTableBase - 0x%Lx\n",
> (UINT64)*PageTableBase
> ));
Will update it in my next version change.
Thanks,
Eric
>
> Thanks
> Laszlo
>
> > + DEBUG ((DEBUG_INFO, " Attributes - 0x%x\n", *Attributes));
> >
> > return ;
> > }
> >
prev parent reply other threads:[~2019-09-12 2:54 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-09-11 1:45 [Patch] UefiCpuPkg/CpuDxe: clean up PAGE_TABLE_LIB_PAGING_CONTEXT usage Dong, Eric
2019-09-11 15:25 ` Laszlo Ersek
2019-09-12 2:54 ` Dong, Eric [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=ED077930C258884BBCB450DB737E662259EE48A5@shsmsx102.ccr.corp.intel.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