Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.kernel > #1516483 > unrolled thread

[PATCH] MIPS: R2-on-R6 emulation bugfix of BLEZL and BGTZL instructions

Started byLeonid Yegoshin <Leonid.Yegoshin@imgtec.com>
First post2016-11-07 20:00 +0100
Last post2016-11-08 12:20 +0100
Articles 2 — 2 participants

Back to article view | Back to linux.kernel


Contents

  [PATCH] MIPS: R2-on-R6 emulation bugfix of BLEZL and BGTZL  instructions Leonid Yegoshin <Leonid.Yegoshin@imgtec.com> - 2016-11-07 20:00 +0100
    Re: [PATCH] MIPS: R2-on-R6 emulation bugfix of BLEZL and BGTZL  instructions "Maciej W. Rozycki" <macro@imgtec.com> - 2016-11-08 12:20 +0100

#1516483 — [PATCH] MIPS: R2-on-R6 emulation bugfix of BLEZL and BGTZL instructions

FromLeonid Yegoshin <Leonid.Yegoshin@imgtec.com>
Date2016-11-07 20:00 +0100
Subject[PATCH] MIPS: R2-on-R6 emulation bugfix of BLEZL and BGTZL instructions
Message-ID<sAXjc-74X-39@gated-at.bofh.it>
MIPS R2 emulation doesn't take into account that BLEZL and BGTZL instructions
require register RT = 0. If it is not zero it can be some legitimate MIPS R6
instruction.

Problem happens after emulation optimization then emulation routine tries
to pipeline emulation and after emulation of one instruction it picks up
a next candidate. In single pass strategy it does not happen because CPU
doesn't trap on branch-compacts which share opcode space with BLEZL/BGTZL
(but has RT != 0, of course).

Signed-off-by: Leonid Yegoshin <Leonid.Yegoshin@imgtec.com>
Reported-by: Douglas Leung <Douglas.Leung@imgtec.com>
---
 arch/mips/kernel/mips-r2-to-r6-emul.c |   11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/arch/mips/kernel/mips-r2-to-r6-emul.c b/arch/mips/kernel/mips-r2-to-r6-emul.c
index 22dedd62818a..b0c86b08c0b9 100644
--- a/arch/mips/kernel/mips-r2-to-r6-emul.c
+++ b/arch/mips/kernel/mips-r2-to-r6-emul.c
@@ -919,6 +919,7 @@ int mipsr2_decoder(struct pt_regs *regs, u32 inst, unsigned long *fcr31)
 		BUG();
 		return SIGEMT;
 	}
+	err = 0;
 	pr_debug("Emulating the 0x%08x R2 instruction @ 0x%08lx (pass=%d))\n",
 		 inst, epc, pass);
 
@@ -1096,10 +1097,16 @@ int mipsr2_decoder(struct pt_regs *regs, u32 inst, unsigned long *fcr31)
 		}
 		break;
 
-	case beql_op:
-	case bnel_op:
 	case blezl_op:
 	case bgtzl_op:
+		/* return MIPS R6 instruction to CPU execution */
+		if (MIPSInst_RT(inst)) {
+			err = SIGILL;
+			break;
+		}
+
+	case beql_op:
+	case bnel_op:
 		if (delay_slot(regs)) {
 			err = SIGILL;
 			break;

[toc] | [next] | [standalone]


#1517088

From"Maciej W. Rozycki" <macro@imgtec.com>
Date2016-11-08 12:20 +0100
Message-ID<sBcBA-bB-27@gated-at.bofh.it>
In reply to#1516483
On Mon, 7 Nov 2016, Leonid Yegoshin wrote:

> MIPS R2 emulation doesn't take into account that BLEZL and BGTZL instructions
> require register RT = 0. If it is not zero it can be some legitimate MIPS R6
> instruction.

 Well, it *is* rather than just can be -- one of BLEZC/BGEZC/BGEC or 
BGTZC/BLTZC/BLTC, respectively, according to the bit patterns in RS/RT, 
all these instructions being compact branches, so we can stop emulation 
rather than decoding them.

 Also please line-wrap your description at 75 columns, as per 
Documentation/SubmittingPatches.

> diff --git a/arch/mips/kernel/mips-r2-to-r6-emul.c b/arch/mips/kernel/mips-r2-to-r6-emul.c
> index 22dedd62818a..b0c86b08c0b9 100644
> --- a/arch/mips/kernel/mips-r2-to-r6-emul.c
> +++ b/arch/mips/kernel/mips-r2-to-r6-emul.c
> @@ -919,6 +919,7 @@ int mipsr2_decoder(struct pt_regs *regs, u32 inst, unsigned long *fcr31)
>  		BUG();
>  		return SIGEMT;
>  	}
> +	err = 0;
>  	pr_debug("Emulating the 0x%08x R2 instruction @ 0x%08lx (pass=%d))\n",
>  		 inst, epc, pass);

 Is this because of BRANCH_LIKELY_TAKEN?  It has to be a separate patch 
then, with a suitable description.

> @@ -1096,10 +1097,16 @@ int mipsr2_decoder(struct pt_regs *regs, u32 inst,
> unsigned long *fcr31)
>  		}
>  		break;
>  
> -	case beql_op:
> -	case bnel_op:
>  	case blezl_op:
>  	case bgtzl_op:
> +		/* return MIPS R6 instruction to CPU execution */
> +		if (MIPSInst_RT(inst)) {
> +			err = SIGILL;
> +			break;
> +		}

 Please add:

		/* Fall through.  */

here so that it is clear it's not a bug; also GCC 7 will catch such cases 
and issue warnings, which I expect according to our settings will cause a 
build failure here if this is missing.

> +
> +	case beql_op:
> +	case bnel_op:

 This part looks fine to me otherwise.

  Maciej

[toc] | [prev] | [standalone]


Back to top | Article view | linux.kernel


csiph-web