All of lore.kernel.org
 help / color / mirror / Atom feed
From: Julien Grall <julien.grall@linaro.org>
To: Ian Campbell <Ian.Campbell@citrix.com>
Cc: stefano.stabellini@eu.citrix.com, tim@xen.org,
	patches@linaro.org, xen-devel@lists.xen.org
Subject: Re: [PATCH v4 1/2] xen/arm: Start to implement an ARM decoder instruction
Date: Thu, 08 Aug 2013 12:52:53 +0100	[thread overview]
Message-ID: <52038695.4080808@linaro.org> (raw)
In-Reply-To: <1375960084.970.68.camel@kazak.uk.xensource.com>

On 08/08/2013 12:08 PM, Ian Campbell wrote:
> On Tue, 2013-08-06 at 19:31 +0100, Julien Grall wrote:
>> Some erratas on ARM processor requires to decode the instruction.
> 
> Errata is already the plural of Erratum, I think.

Right, I will fix it.

> 
>> The decoder will, obviously, decode and fill the ISS fields of the hsr_dabt.
>>
>> For the moment, the decoder only supports:
>>     - THUMB2 store instruction
>>     - THUMB single load/store instruction
>>
>> Signed-off-by: Julien Grall <julien.grall@linaro.org>
>>
>> ---
>>     Changes in v4:
>>         - Add warning about the purpose of this function
>>         - Add helper to update ISS (register, sign, size) field of DABT
>>         - Improve decoding for THUMB 2 store instruction
>>         - Only decode thumb instruction if it's a 32-bit guest
>> ---
>>  xen/arch/arm/Makefile |    1 +
>>  xen/arch/arm/decode.c |  168 +++++++++++++++++++++++++++++++++++++++++++++++++
>>  xen/arch/arm/decode.h |   49 +++++++++++++++
>>  3 files changed, 218 insertions(+)
>>  create mode 100644 xen/arch/arm/decode.c
>>  create mode 100644 xen/arch/arm/decode.h
>>
>> diff --git a/xen/arch/arm/Makefile b/xen/arch/arm/Makefile
>> index 5ae5831..5c13a65 100644
>> --- a/xen/arch/arm/Makefile
>> +++ b/xen/arch/arm/Makefile
>> @@ -30,6 +30,7 @@ obj-y += vtimer.o
>>  obj-y += vpl011.o
>>  obj-y += hvm.o
>>  obj-y += device.o
>> +obj-y += decode.o
>>  
>>  #obj-bin-y += ....o
>>  
>> diff --git a/xen/arch/arm/decode.c b/xen/arch/arm/decode.c
>> new file mode 100644
>> index 0000000..0e7d8ba
>> --- /dev/null
>> +++ b/xen/arch/arm/decode.c
>> @@ -0,0 +1,168 @@
>> +/*
>> + * xen/arch/arm/decode.c
>> + *
>> + * Instruction decoder
>> + *
>> + * Julien Grall <julien.grall@linaro.org>
>> + * Copyright (C) 2013 Linaro Limited.
>> + *
>> + * This program is free software; you can redistribute it and/or modify
>> + * it under the terms of the GNU General Public License as published by
>> + * the Free Software Foundation; either version 2 of the License, or
>> + * (at your option) any later version.
>> + *
>> + * This program is distributed in the hope that it will be useful,
>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
>> + * GNU General Public License for more details.
>> + */
>> +
>> +#include <xen/types.h>
>> +#include <xen/sched.h>
>> +#include <asm/current.h>
>> +#include <asm/guest_access.h>
>> +#include <xen/lib.h>
>> +
>> +#include "decode.h"
>> +
>> +static void update_dabt(struct hsr_dabt *dabt, int reg,
>> +                        uint8_t size, bool_t sign)
>> +{
>> +    dabt->reg = reg;
>> +    dabt->size = size;
>> +    dabt->sign = sign;
>> +}
>> +
>> +/* TODO: Handle all THUMB2 instruction other than simple store */
> 
> I think we decided to remove this (and the other one).

Oh right, I forgot to remove it.

>> +static int decode_thumb2(register_t pc, struct hsr_dabt *dabt, uint16_t hw1)
>> +{
>> +    uint16_t hw2;
>> +    int rc;
>> +    uint16_t rt;
>> +
>> +    rc = raw_copy_from_guest(&hw2, (void *__user)(pc + 2), sizeof (hw2));
>> +    if ( rc )
>> +        return rc;
>> +
>> +    rt = (hw2 >> 12) & 0x7;
>> +
>> +    switch ( (hw1 >> 9) & 0xf )
>> +    {
>> +    case 12:
>> +    {
>> +        bool_t sign = !!(hw1 & (1 << 8));
>> +        bool_t load = !!(hw1 & (1 << 4));
> 
> !! is unneeded with the bool type I think.

I was not sure, I will remove the '!!'.

>> +
>> +        if ( (hw1 & 0x0110) == 0x0100 )
>> +            /* NEON instruction */
>> +            goto bad_thumb2;
>> +
>> +        if ( (hw1 & 0x0070) == 0x0070 )
>> +            /* Undefined opcodes */
>> +            goto bad_thumb2;
>> +
>> +        /* Store/Load single data item */
>> +        if ( rt == 15 )
>> +            /* XXX: Rt == 15 is only invalid for store instruction */
> 
> So the check should be "rt == 15 && load" ?

No, when rt == 15 the instructions are memory hints (pld, pldw,..) and
the layout is different.

>> +            goto bad_thumb2;
>> +
>> +        if ( !load && sign )
>> +            /* Store instruction doesn't support sign extension */
>> +            goto bad_thumb2;
>> +
>> +        update_dabt(dabt, rt, (hw1 >> 5) & 3, sign);
>> +
>> +        break;
>> +    }
>> +    default:
>> +        goto bad_thumb2;
>> +    }
>> +
>> +    return 0;
>> +
>> +bad_thumb2:
>> +    printk("DOM%u: unhandled THUMB2 instruction 0x%x%x\n",
>> +           current->domain->domain_id, hw1, hw2);
> 
> Should this be gdprintk or gprintk?

I will use gdprintk.

> 
>> +
>> +    return 1;
>> +}
>> +
>> +/* TODO: Handle all THUMB instructions other than store */
>> +static int decode_thumb(register_t pc, struct hsr_dabt *dabt)
>> +{
>> +    uint16_t instr;
>> +    int rc;
>> +
>> +    rc = raw_copy_from_guest(&instr, (void * __user)pc, sizeof (instr));
>> +    if ( rc )
>> +        return rc;
>> +
>> +    switch ( instr >> 12 )
>> +    {
>> +    case 5:
>> +    {
>> +        /* Load/Store register */
>> +        uint16_t opB = (instr >> 9) & 0x7;
>> +        int reg = instr & 7;
>> +
>> +        switch ( opB & 0x3 )
>> +        {
>> +        case 0: /* Non-signed word */
>> +            update_dabt(dabt, reg, 2, 0);
>> +            break;
>> +        case 1: /* Non-signed halfword */
>> +            update_dabt(dabt, reg, 1, 0);
>> +            break;
>> +        case 2: /* Non-signed byte */
>> +            update_dabt(dabt, reg, 0, 0);
>> +            break;
>> +        case 3: /* Signed byte */
>> +            update_dabt(dabt, reg, 0, 1);
>> +            break;
>> +        }
>> +
>> +        break;
>> +    }
>> +    case 6:
>> +        /* Load/Store word immediate offset */
>> +        update_dabt(dabt, instr & 7, 2, 0);
>> +        break;
>> +    case 7:
>> +        /* Load/Store byte immediate offset */
>> +        update_dabt(dabt, instr & 7, 0, 0);
>> +        break;
>> +    case 8:
>> +        /* Load/Store halfword immediate offset */
>> +        update_dabt(dabt, instr & 7, 1, 0);
>> +        break;
>> +    case 9:
>> +        /* Load/Store word sp offset */
>> +        update_dabt(dabt, (instr >> 8) & 7, 2, 0);
>> +        break;
>> +    case 14:
>> +        if ( instr & (1 << 11) )
>> +            return decode_thumb2(pc, dabt, instr);
>> +        goto bad_thumb;
>> +    case 15:
>> +        return decode_thumb2(pc, dabt, instr);
>> +    default:
>> +        goto bad_thumb;
>> +    }
>> +
>> +    return 0;
>> +
>> +bad_thumb:
>> +    printk("DOM%u: unhandled THUMB instruction 0x%x\n",
>> +           current->domain->domain_id, instr);
> 
> gdprint etc?
> 
>> +    return 1;
>> +}
>> +
>> +int decode_instruction(const struct cpu_user_regs *regs, struct hsr_dabt *dabt)
>> +{
>> +    if ( is_pv32_domain(current->domain) && regs->cpsr & PSR_THUMB )
>> +        return decode_thumb(regs->pc, dabt);
>> +
>> +    /* TODO: Handle ARM instruction */
> 
> You log unhandled thumb instructions but not ARM one, I think you should
> log here too.

I'll add a warning.

>> +
>> +    return 1;
>> +}
>> diff --git a/xen/arch/arm/decode.h b/xen/arch/arm/decode.h
>> new file mode 100644
>> index 0000000..4613763
>> --- /dev/null
>> +++ b/xen/arch/arm/decode.h
>> @@ -0,0 +1,49 @@
>> +/*
>> + * xen/arch/arm/decode.h
>> + *
>> + * Instruction decoder
>> + *
>> + * Julien Grall <julien.grall@linaro.org>
>> + * Copyright (C) 2013 Linaro Limited.
>> + *
>> + * This program is free software; you can redistribute it and/or modify
>> + * it under the terms of the GNU General Public License as published by
>> + * the Free Software Foundation; either version 2 of the License, or
>> + * (at your option) any later version.
>> + *
>> + * This program is distributed in the hope that it will be useful,
>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
>> + * GNU General Public License for more details.
>> + */
>> +
>> +#ifndef __ARCH_ARM_DECODE_H_
>> +#define __ARCH_ARM_DECODE_H_
>> +
>> +#include <asm/regs.h>
>> +#include <asm/processor.h>
>> +
>> +/**
>> + * Decode an instruction from pc
>> + * /!\ This function is not intended to fully decode an instruction. It
>> + * considers that the instruction is valid.
>> + *
>> + * This function will get:
>> + *  - The transfer register
>> + *  - Sign bit
>> + *  - Size
>> + */
>> +
>> +int decode_instruction(const struct cpu_user_regs *regs,
>> +                       struct hsr_dabt *dabt);
>> +
>> +#endif /* __ARCH_ARM_DECODE_H_ */
>> +
>> +/*
>> + * Local variables:
>> + * mode: C
>> + * c-file-style: "BSD"
>> + * c-basic-offset: 4
>> + * indent-tabs-mode: nil
>> + * End:
>> + */
> 
> 


-- 
Julien Grall

  reply	other threads:[~2013-08-08 11:52 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-08-06 18:31 [PATCH v4 0/2] Implement errata 766422 Julien Grall
2013-08-06 18:31 ` [PATCH v4 1/2] xen/arm: Start to implement an ARM decoder instruction Julien Grall
2013-08-08 11:08   ` Ian Campbell
2013-08-08 11:52     ` Julien Grall [this message]
2013-08-08 12:13     ` Julien Grall
2013-08-06 18:31 ` [PATCH v4 2/2] xen/arm: errata 766422: decode thumb store during data abort Julien Grall
2013-08-08 11:09   ` Ian Campbell
2013-08-08 12:04     ` Julien Grall
2013-08-08 12:12       ` Ian Campbell

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-all 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=52038695.4080808@linaro.org \
    --to=julien.grall@linaro.org \
    --cc=Ian.Campbell@citrix.com \
    --cc=patches@linaro.org \
    --cc=stefano.stabellini@eu.citrix.com \
    --cc=tim@xen.org \
    --cc=xen-devel@lists.xen.org \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.