xen-devel.lists.xenproject.org archive mirror
 help / color / mirror / Atom feed
From: Julien Grall <julien@xen.org>
To: Michal Orzel <michal.orzel@arm.com>, xen-devel@lists.xenproject.org
Cc: Stefano Stabellini <sstabellini@kernel.org>,
	Volodymyr Babchuk <Volodymyr_Babchuk@epam.com>,
	bertrand.marquis@arm.com
Subject: Re: [PATCH 9/9] xen/arm64: Remove READ/WRITE_SYSREG32 helper macros
Date: Tue, 27 Apr 2021 09:30:59 +0100	[thread overview]
Message-ID: <0d4ffc04-3fbb-4b62-f6d8-e33da427e505@xen.org> (raw)
In-Reply-To: <a37f7d85-dee7-d2f8-9883-1a8004d1ca25@arm.com>



On 27/04/2021 08:16, Michal Orzel wrote:
> Hi Jilien,

Hi,

> On 21.04.2021 21:16, Julien Grall wrote:
>> Hi Michal,
>>
>> On 20/04/2021 08:08, Michal Orzel wrote:
>>> AArch64 system registers are 64bit whereas AArch32 ones
>>> are 32bit or 64bit. MSR/MRS are expecting 64bit values thus
>>> we should get rid of helpers READ/WRITE_SYSREG32
>>> in favour of using READ/WRITE_SYSREG.
>>> We should also use register_t type when reading sysregs
>>> which can correspond to uint64_t or uint32_t.
>>> Even though many AArch64 sysregs have upper 32bit reserved
>>> it does not mean that they can't be widen in the future.
>>
>> As a general comment, all your commit message contains information about the goal (which is great). But they don't go much in details about the actual changes. I have tried to point out where I think those details would be helpful.
>>
>>>
>>> As there are no other places in the code using READ/WRITE_SYSREG32,
>>> remove the helper macros.
>>>
>>> Signed-off-by: Michal Orzel <michal.orzel@arm.com>
>>> ---
>>>    xen/arch/arm/vcpreg.c               | 16 ++++++++++++++++
>>>    xen/include/asm-arm/arm64/sysregs.h |  5 -----
>>>    2 files changed, 16 insertions(+), 5 deletions(-)
>>>
>>> diff --git a/xen/arch/arm/vcpreg.c b/xen/arch/arm/vcpreg.c
>>> index c7f516ee0a..6cb7f3108c 100644
>>> --- a/xen/arch/arm/vcpreg.c
>>> +++ b/xen/arch/arm/vcpreg.c
>>> @@ -48,6 +48,7 @@
>>>     */
>>>      /* The name is passed from the upper macro to workaround macro expansion. */
>>> +#ifdef CONFIG_ARM_32
>>>    #define TVM_REG(sz, func, reg...)                                           \
>>>    static bool func(struct cpu_user_regs *regs, uint##sz##_t *r, bool read)    \
>>>    {                                                                           \
>>> @@ -61,6 +62,21 @@ static bool func(struct cpu_user_regs *regs, uint##sz##_t *r, bool read)    \
>>>                                                                                \
>>>        return true;                                                            \
>>>    }
>>> +#else /* CONFIG_ARM_64 */
>>> +#define TVM_REG(sz, func, reg...)                                           \
>>> +static bool func(struct cpu_user_regs *regs, uint##sz##_t *r, bool read)    \
>>> +{                                                                           \
>>> +    struct vcpu *v = current;                                               \
>>> +    bool cache_enabled = vcpu_has_cache_enabled(v);                         \
>>> +                                                                            \
>>> +    GUEST_BUG_ON(read);                                                     \
>>> +    WRITE_SYSREG(*r, reg);                                                  \
>>> +                                                                            \
>>> +    p2m_toggle_cache(v, cache_enabled);                                     \
>>> +                                                                            \
>>> +    return true;                                                            \
>>> +}
>>> +#endif
>>
>> It would be nice if this change can be explained in the commit message. However, I think we can avoid the duplication by aliasing TVM_REG32() to TVM_REG64() on Arm64.
>>
> Unfortunately we cannot. This is the only working solution for now.
> If we do the alias we will get many errors due to incompatbile pointer type in vreg_emulate_*.
> Without a proper change in this functions we can't do that.

Right, so the prototype needs to stay the same. How about provide a 
wrapper WRITE_SYSREG_SZ(sz, val, sysreg) internal to vcpreg.c?

On 32-bit it would expand to WRITE_SYSREG##sz(val, sysreg) and on 
64-bit, it would expand to WRITE_SYSREG().

> I will modify it once I start working on vreg_emulate topic but it requires more investigation.

While it would be great to get rid of {READ, WRITE}_SYSREG32, I don't 
want to duplicate the code (even temporarily) just for the sake for 
removing the helpers.

If the new proposal I suggested above doesn't work, then I would 
strongly prefer to keep the existing code until vreg_emulate_* is reworked.

Cheers,

-- 
Julien Grall


      reply	other threads:[~2021-04-27  8:31 UTC|newest]

Thread overview: 27+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-04-20  7:08 [PATCH 0/9] xen/arm64: Get rid of READ/WRITE_SYSREG32 Michal Orzel
2021-04-20  7:08 ` [PATCH 1/9] arm64/vfp: " Michal Orzel
2021-04-20 13:04   ` Julien Grall
2021-04-20  7:08 ` [PATCH 2/9] arm/domain: " Michal Orzel
2021-04-20 13:12   ` Julien Grall
2021-04-21  7:36     ` Michal Orzel
2021-04-21 10:20       ` Julien Grall
2021-04-20  7:08 ` [PATCH 3/9] arm/gic: " Michal Orzel
2021-04-20 13:28   ` Julien Grall
2021-04-21  7:48     ` Michal Orzel
2021-04-21 10:21       ` Julien Grall
2021-04-20  7:08 ` [PATCH 4/9] arm/p2m: " Michal Orzel
2021-04-20 13:31   ` Julien Grall
2021-04-20  7:08 ` [PATCH 5/9] arm/mm: " Michal Orzel
2021-04-20 13:37   ` Julien Grall
2021-04-22 11:47     ` Michal Orzel
2021-04-22 13:40       ` Julien Grall
2021-04-20  7:08 ` [PATCH 6/9] arm/page: " Michal Orzel
2021-04-21 15:11   ` Julien Grall
2021-04-20  7:08 ` [PATCH 7/9] arm/time,vtimer: " Michal Orzel
2021-04-21 16:01   ` Julien Grall
2021-04-20  7:08 ` [PATCH 8/9] arm: Change type of hsr to register_t Michal Orzel
2021-04-21 19:02   ` Julien Grall
2021-04-20  7:08 ` [PATCH 9/9] xen/arm64: Remove READ/WRITE_SYSREG32 helper macros Michal Orzel
2021-04-21 19:16   ` Julien Grall
2021-04-27  7:16     ` Michal Orzel
2021-04-27  8:30       ` Julien Grall [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-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=0d4ffc04-3fbb-4b62-f6d8-e33da427e505@xen.org \
    --to=julien@xen.org \
    --cc=Volodymyr_Babchuk@epam.com \
    --cc=bertrand.marquis@arm.com \
    --cc=michal.orzel@arm.com \
    --cc=sstabellini@kernel.org \
    --cc=xen-devel@lists.xenproject.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).