From: "Ard Biesheuvel" <ard.biesheuvel@linaro.org>
To: Leif Lindholm <leif.lindholm@linaro.org>
Cc: Baptiste Gerondeau <baptiste.gerondeau@linaro.org>,
edk2-devel-groups-io <devel@edk2.groups.io>,
"Kinney, Michael D" <michael.d.kinney@intel.com>,
"Gao, Liming" <liming.gao@intel.com>,
"Zhang, Shenglei" <shenglei.zhang@intel.com>,
Baptiste GERONDEAU <bgerondeau@gmail.com>
Subject: Re: [PATCH 2/3] ARM/Assembler: Correct syntax from RVCT for MSFT
Date: Thu, 19 Sep 2019 13:01:04 +0300 [thread overview]
Message-ID: <CAKv+Gu_-mwE2Qa96i4mX5UbgXrfFHC3pC28TOQcSncw8DJUa8g@mail.gmail.com> (raw)
In-Reply-To: <20190919094846.GO28454@bivouac.eciton.net>
On Thu, 19 Sep 2019 at 12:48, Leif Lindholm <leif.lindholm@linaro.org> wrote:
>
> On Thu, Sep 19, 2019 at 12:32:56PM +0300, Ard Biesheuvel wrote:
> > On Wed, 18 Sep 2019 at 15:27, Baptiste Gerondeau
> > <baptiste.gerondeau@linaro.org> wrote:
> > >
> > > From: Baptiste GERONDEAU <bgerondeau@gmail.com>
> > >
> > > RVCT and MSFT's ARM assembler share the same file syntax, but some
> > > instructions use pre-UAL syntax that is not picked up
> > > by MSFT's ARM assembler, this commit translates those instructions
> > > into MSFT-buildable ones (subset of UAL/THUMB).
> > >
> > > Signed-off-by: Baptiste Gerondeau <baptiste.gerondeau@linaro.org>
> > > ---
> > > ArmPkg/Library/ArmExceptionLib/Arm/ExceptionSupport.asm | 30 +++++++++++++++++-------------
> > > ArmPkg/Library/ArmLib/Arm/ArmV7Support.asm | 6 ++++--
> > > MdePkg/Library/BaseIoLibIntrinsic/Arm/ArmVirtMmio.asm | 18 +++++++++---------
> > > 3 files changed, 30 insertions(+), 24 deletions(-)
> > >
> > > diff --git a/ArmPkg/Library/ArmExceptionLib/Arm/ExceptionSupport.asm b/ArmPkg/Library/ArmExceptionLib/Arm/ExceptionSupport.asm
> > > index aa0229d2e85f..880246bd6206 100644
> > > --- a/ArmPkg/Library/ArmExceptionLib/Arm/ExceptionSupport.asm
> > > +++ b/ArmPkg/Library/ArmExceptionLib/Arm/ExceptionSupport.asm
> > > @@ -90,7 +90,7 @@ Fiq
> > > ResetEntry
> > > srsfd #0x13! ; Store return state on SVC stack
> > > ; We are already in SVC mode
> > > - stmfd SP!,{LR} ; Store the link register for the current mode
> > > + push {LR} ; Store the link register for the current mode
> > > sub SP,SP,#0x20 ; Save space for SP, LR, PC, IFAR - CPSR
> > > stmfd SP!,{R0-R12} ; Store the register state
> > >
> > > @@ -102,7 +102,7 @@ UndefinedInstructionEntry
> > > sub LR, LR, #4 ; Only -2 for Thumb, adjust in CommonExceptionEntry
> > > srsfd #0x13! ; Store return state on SVC stack
> > > cps #0x13 ; Switch to SVC for common stack
> > > - stmfd SP!,{LR} ; Store the link register for the current mode
> > > + push {LR} ; Store the link register for the current mode
> > > sub SP,SP,#0x20 ; Save space for SP, LR, PC, IFAR - CPSR
> > > stmfd SP!,{R0-R12} ; Store the register state
> > >
> > > @@ -113,7 +113,7 @@ UndefinedInstructionEntry
> > > SoftwareInterruptEntry
> > > srsfd #0x13! ; Store return state on SVC stack
> > > ; We are already in SVC mode
> > > - stmfd SP!,{LR} ; Store the link register for the current mode
> > > + push {LR} ; Store the link register for the current mode
> > > sub SP,SP,#0x20 ; Save space for SP, LR, PC, IFAR - CPSR
> > > stmfd SP!,{R0-R12} ; Store the register state
> > >
> > > @@ -125,7 +125,7 @@ PrefetchAbortEntry
> > > sub LR,LR,#4
> > > srsfd #0x13! ; Store return state on SVC stack
> > > cps #0x13 ; Switch to SVC for common stack
> > > - stmfd SP!,{LR} ; Store the link register for the current mode
> > > + push {LR} ; Store the link register for the current mode
> > > sub SP,SP,#0x20 ; Save space for SP, LR, PC, IFAR - CPSR
> > > stmfd SP!,{R0-R12} ; Store the register state
> > >
> > > @@ -137,7 +137,7 @@ DataAbortEntry
> > > sub LR,LR,#8
> > > srsfd #0x13! ; Store return state on SVC stack
> > > cps #0x13 ; Switch to SVC for common stack
> > > - stmfd SP!,{LR} ; Store the link register for the current mode
> > > + push {LR} ; Store the link register for the current mode
> > > sub SP,SP,#0x20 ; Save space for SP, LR, PC, IFAR - CPSR
> > > stmfd SP!,{R0-R12} ; Store the register state
> > >
> > > @@ -148,7 +148,7 @@ DataAbortEntry
> > > ReservedExceptionEntry
> > > srsfd #0x13! ; Store return state on SVC stack
> > > cps #0x13 ; Switch to SVC for common stack
> > > - stmfd SP!,{LR} ; Store the link register for the current mode
> > > + push {LR} ; Store the link register for the current mode
> > > sub SP,SP,#0x20 ; Save space for SP, LR, PC, IFAR - CPSR
> > > stmfd SP!,{R0-R12} ; Store the register state
> > >
> > > @@ -160,7 +160,7 @@ IrqEntry
> > > sub LR,LR,#4
> > > srsfd #0x13! ; Store return state on SVC stack
> > > cps #0x13 ; Switch to SVC for common stack
> > > - stmfd SP!,{LR} ; Store the link register for the current mode
> > > + push {LR} ; Store the link register for the current mode
> > > sub SP,SP,#0x20 ; Save space for SP, LR, PC, IFAR - CPSR
> > > stmfd SP!,{R0-R12} ; Store the register state
> > >
> > > @@ -172,7 +172,7 @@ FiqEntry
> > > sub LR,LR,#4
> > > srsfd #0x13! ; Store return state on SVC stack
> > > cps #0x13 ; Switch to SVC for common stack
> > > - stmfd SP!,{LR} ; Store the link register for the current mode
> > > + push {LR} ; Store the link register for the current mode
> > > sub SP,SP,#0x20 ; Save space for SP, LR, PC, IFAR - CPSR
> > > stmfd SP!,{R0-R12} ; Store the register state
> > > ; Since we have already switch to SVC R8_fiq - R12_fiq
> > > @@ -213,9 +213,11 @@ AsmCommonExceptionEntry
> > > and R3, R1, #0x1f ; Check CPSR to see if User or System Mode
> > > cmp R3, #0x1f ; if ((CPSR == 0x10) || (CPSR == 0x1f))
> > > cmpne R3, #0x10 ;
> > > - stmeqed R2, {lr}^ ; save unbanked lr
> > > + mrseq R8, lr_usr ; save unbanked lr to R8
> > > + streq R2, [R8] ; make R2 point to R8
> > > ; else
> > > - stmneed R2, {lr} ; save SVC lr
> > > + mrsne R8, lr_svc ; save SVC lr to R8
> > > + strne R2, [R8] ; make R2 point to R8
> > >
> >
> > Can you make this str unconditional, and drop the former?
>
> Yeah, that would be an improvement.
>
> > >
> > > ldr R5, [SP, #0x58] ; PC is the LR pushed by srsfd
> > > @@ -280,15 +282,17 @@ CommonCExceptionHandler (
> > > and R1, R1, #0x1f ; Check to see if User or System Mode
> > > cmp R1, #0x1f ; if ((CPSR == 0x10) || (CPSR == 0x1f))
> > > cmpne R1, #0x10 ;
> > > - ldmeqed R2, {lr}^ ; restore unbanked lr
> > > + ldreq R8, [R2] ; load sys/usr lr from R2 pointer
> > > + msreq lr_usr, R8 ; restore unbanked lr
> > > ; else
> > > - ldmneed R3, {lr} ; restore SVC lr, via ldmfd SP!, {LR}
> > > + ldrne R8, [R3] ; load SVC lr from R3 pointer
> > > + msrne lr_svc, R8 ; restore SVC lr, via ldmfd SP!, {LR}
> > >
> > > ldmfd SP!,{R0-R12} ; Restore general purpose registers
> > > ; Exception handler can not change SP
> > >
> > > add SP,SP,#0x20 ; Clear out the remaining stack space
> > > - ldmfd SP!,{LR} ; restore the link register for this context
> > > + pop {LR} ; restore the link register for this context
> > > rfefd SP! ; return from exception via srsfd stack slot
> > >
> > > END
> > > diff --git a/ArmPkg/Library/ArmLib/Arm/ArmV7Support.asm b/ArmPkg/Library/ArmLib/Arm/ArmV7Support.asm
> > > index 3146c2b52181..724306399e6c 100644
> > > --- a/ArmPkg/Library/ArmLib/Arm/ArmV7Support.asm
> > > +++ b/ArmPkg/Library/ArmLib/Arm/ArmV7Support.asm
> > > @@ -200,8 +200,10 @@ Loop2
> > > mov R9, R4 ; R9 working copy of the max way size (right aligned)
> > >
> > > Loop3
> > > - orr R0, R10, R9, LSL R5 ; factor in the way number and cache number into R11
> > > - orr R0, R0, R7, LSL R2 ; factor in the index number
> > > + lsl R8, R9, R5
> > > + orr R0, R10, R8 ; factor in the way number and cache number
> > > + lsl R8, R7, R2
> > > + orr R0, R0, R8 ; factor in the index number
> > >
> >
> > What's wrong with this code?
>
> Inline barrel shifter is only available in A32, not T32 (and VS seems
> to love T32).
>
So is this simply the default of the compiler? I'd prefer it if we
could add a 'CODE 32' directive instead, that way, we may not need any
of the other changes to begin with.
next prev parent reply other threads:[~2019-09-19 10:01 UTC|newest]
Thread overview: 32+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-09-18 12:25 [PATCH 0/3] Arm builds on Visual Studio Baptiste Gerondeau
2019-09-18 12:25 ` [PATCH 1/3] ArmPkg/MdePkg : Unify INF files format Baptiste Gerondeau
2019-09-19 9:29 ` Ard Biesheuvel
2019-09-18 12:25 ` [PATCH 2/3] ARM/Assembler: Correct syntax from RVCT for MSFT Baptiste Gerondeau
2019-09-19 9:32 ` Ard Biesheuvel
2019-09-19 9:48 ` Leif Lindholm
2019-09-19 10:01 ` Ard Biesheuvel [this message]
2019-09-19 10:09 ` Leif Lindholm
2019-09-19 10:25 ` Ard Biesheuvel
2019-09-19 10:34 ` Baptiste Gerondeau
2019-09-19 10:37 ` Ard Biesheuvel
2019-09-19 10:47 ` Leif Lindholm
2019-09-19 10:53 ` Ard Biesheuvel
2019-09-19 11:25 ` Leif Lindholm
2019-09-19 12:36 ` Ard Biesheuvel
2019-09-19 14:31 ` Leif Lindholm
2019-09-19 14:44 ` Ard Biesheuvel
2019-09-19 11:07 ` Baptiste Gerondeau
2019-09-19 10:37 ` Leif Lindholm
2019-09-18 12:25 ` [PATCH 3/3] ARM/Assembler: Reuse RVCT assembler for MSFT build Baptiste Gerondeau
2019-09-19 9:38 ` Ard Biesheuvel
2019-09-19 9:52 ` Leif Lindholm
2019-09-19 9:59 ` Ard Biesheuvel
2019-09-18 16:43 ` [PATCH 0/3] Arm builds on Visual Studio Leif Lindholm
2019-09-19 6:19 ` Liming Gao
2019-09-19 9:44 ` Leif Lindholm
2019-09-19 14:53 ` Liming Gao
2019-09-19 19:24 ` [edk2-devel] " Laszlo Ersek
2019-09-19 19:57 ` Andrew Fish
2019-09-19 20:27 ` Leif Lindholm
2019-09-24 1:28 ` Liming Gao
[not found] <cover.1568821123.git.baptiste.gerondeau@linaro.org>
2019-09-18 16:05 ` [PATCH 2/3] ARM/Assembler: Correct syntax from RVCT for MSFT Baptiste Gerondeau
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=CAKv+Gu_-mwE2Qa96i4mX5UbgXrfFHC3pC28TOQcSncw8DJUa8g@mail.gmail.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