All of lore.kernel.org
 help / color / mirror / Atom feed
From: Jonathan Cameron <Jonathan.Cameron@huawei.com>
To: "Philippe Mathieu-Daudé" <philmd@linaro.org>
Cc: <qemu-devel@nongnu.org>, Michael Tsirkin <mst@redhat.com>,
	Fan Ni <fan.ni@samsung.com>, <linux-cxl@vger.kernel.org>,
	Dave Jiang <dave.jiang@intel.com>, <linuxarm@huawei.com>
Subject: Re: [PATCH 2/2] hw/cxl: Support 4 HDM decoders at all levels of topology
Date: Tue, 5 Sep 2023 16:44:13 +0100	[thread overview]
Message-ID: <20230905164413.0000406c@huawei.com> (raw)
In-Reply-To: <56291b02-5474-77b6-5563-6367bf5dcb4c@linaro.org>

On Mon, 4 Sep 2023 20:36:02 +0200
Philippe Mathieu-Daudé <philmd@linaro.org> wrote:

> Hi Jonathan,
> 
> Few style comments inlined.
> 
> On 4/9/23 18:47, Jonathan Cameron wrote:
> > Support these decoders in CXL host bridges (pxb-cxl), CXL Switch USP
> > and CXL Type 3 end points.
> > 
> > Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
> > ---
Hi Philippe,

Thanks for the particularly quick reviews! 

...

> > diff --git a/hw/cxl/cxl-component-utils.c b/hw/cxl/cxl-component-utils.c
> > index e96398e8af..79b9369756 100644
> > --- a/hw/cxl/cxl-component-utils.c
> > +++ b/hw/cxl/cxl-component-utils.c
> > @@ -42,6 +42,9 @@ static void dumb_hdm_handler(CXLComponentState *cxl_cstate, hwaddr offset,
> >   
> >       switch (offset) {
> >       case A_CXL_HDM_DECODER0_CTRL:
> > +    case A_CXL_HDM_DECODER1_CTRL:
> > +    case A_CXL_HDM_DECODER2_CTRL:
> > +    case A_CXL_HDM_DECODER3_CTRL:
> >           should_commit = FIELD_EX32(value, CXL_HDM_DECODER0_CTRL, COMMIT);
> >           should_uncommit = !should_commit;
> >           break;
> > @@ -81,7 +84,7 @@ static void cxl_cache_mem_write_reg(void *opaque, hwaddr offset, uint64_t value,
> >       }
> >   
> >       if (offset >= A_CXL_HDM_DECODER_CAPABILITY &&
> > -        offset <= A_CXL_HDM_DECODER0_TARGET_LIST_HI) {
> > +        offset <= A_CXL_HDM_DECODER3_TARGET_LIST_HI) {
> >           dumb_hdm_handler(cxl_cstate, offset, value);
> >       } else {
> >           cregs->cache_mem_registers[offset / sizeof(*cregs->cache_mem_registers)] = value;
> > @@ -161,7 +164,7 @@ static void ras_init_common(uint32_t *reg_state, uint32_t *write_msk)
> >   static void hdm_init_common(uint32_t *reg_state, uint32_t *write_msk,
> >                               enum reg_type type)
> >   {
> > -    int decoder_count = 1;
> > +    int decoder_count = 4;  
> 
>    unsigned decoder_count = HDM_DECODER_COUNT;
> 
> >       int i;
> >   
> >       ARRAY_FIELD_DP32(reg_state, CXL_HDM_DECODER_CAPABILITY, DECODER_COUNT,
> > @@ -174,19 +177,22 @@ static void hdm_init_common(uint32_t *reg_state, uint32_t *write_msk,
> >                        HDM_DECODER_ENABLE, 0);
> >       write_msk[R_CXL_HDM_DECODER_GLOBAL_CONTROL] = 0x3;
> >       for (i = 0; i < decoder_count; i++) {  
> 
> Alternatively:
> 
>          for (i = 0; i < decoder_count; i++, write_msk += 8) {
>              write_msk[R_CXL_HDM_DECODER0_BASE_LO] = 0xf0000000;

That's a bit nasty and fragile given we are offsetting the base register than
indexing into it (so applying a later offset).

> 
> > -        write_msk[R_CXL_HDM_DECODER0_BASE_LO + i * 0x20] = 0xf0000000;
> > -        write_msk[R_CXL_HDM_DECODER0_BASE_HI + i * 0x20] = 0xffffffff;
> > -        write_msk[R_CXL_HDM_DECODER0_SIZE_LO + i * 0x20] = 0xf0000000;
> > -        write_msk[R_CXL_HDM_DECODER0_SIZE_HI + i * 0x20] = 0xffffffff;
> > -        write_msk[R_CXL_HDM_DECODER0_CTRL + i * 0x20] = 0x13ff;
> > +        write_msk[R_CXL_HDM_DECODER0_BASE_LO + i * 0x20 / 4] = 0xf0000000;  
> 
> (this 0x20 / 4 bugs me a bit).

Instead, I've gone with a local variable which leaves me room for deriving
this based on the step between the registers for decoders 0 and 1.

hdm_inc = R_CXL_HDM_DECODER1_BASE_LO - R_CXL_HDM_DECODER0_BASE_LO;

I haven't added a define for this because it would probably have to be
long enough that it will cause line length problems :(
So it is replicated in a few different places which isn't ideal
but definitely better than the 0x20 / 4

> 
> > +        write_msk[R_CXL_HDM_DECODER0_BASE_HI + i * 0x20 / 4]  = 0xffffffff;
> > +        write_msk[R_CXL_HDM_DECODER0_SIZE_LO + i * 0x20 / 4] = 0xf0000000;
> > +        write_msk[R_CXL_HDM_DECODER0_SIZE_HI + i * 0x20 / 4] = 0xffffffff;
> > +        write_msk[R_CXL_HDM_DECODER0_CTRL + i * 0x20 / 4] = 0x13ff;
> >           if (type == CXL2_DEVICE ||
> >               type == CXL2_TYPE3_DEVICE ||
> >               type == CXL2_LOGICAL_DEVICE) {
> > -            write_msk[R_CXL_HDM_DECODER0_TARGET_LIST_LO + i * 0x20] = 0xf0000000;
> > +            write_msk[R_CXL_HDM_DECODER0_TARGET_LIST_LO + i * 0x20 / 4] =
> > +                0xf0000000;
> >           } else {
> > -            write_msk[R_CXL_HDM_DECODER0_TARGET_LIST_LO + i * 0x20] = 0xffffffff;
> > +            write_msk[R_CXL_HDM_DECODER0_TARGET_LIST_LO + i * 0x20 / 4] =
> > +                0xffffffff;
> >           }
> > -        write_msk[R_CXL_HDM_DECODER0_TARGET_LIST_HI + i * 0x20] = 0xffffffff;
> > +        write_msk[R_CXL_HDM_DECODER0_TARGET_LIST_HI + i * 0x20 / 4] =
> > +            0xffffffff;
> >       }
> >   }  
> 


WARNING: multiple messages have this Message-ID (diff)
From: Jonathan Cameron via <qemu-devel@nongnu.org>
To: "Philippe Mathieu-Daudé" <philmd@linaro.org>
Cc: <qemu-devel@nongnu.org>, Michael Tsirkin <mst@redhat.com>,
	Fan Ni <fan.ni@samsung.com>, <linux-cxl@vger.kernel.org>,
	Dave Jiang <dave.jiang@intel.com>, <linuxarm@huawei.com>
Subject: Re: [PATCH 2/2] hw/cxl: Support 4 HDM decoders at all levels of topology
Date: Tue, 5 Sep 2023 16:44:13 +0100	[thread overview]
Message-ID: <20230905164413.0000406c@huawei.com> (raw)
In-Reply-To: <56291b02-5474-77b6-5563-6367bf5dcb4c@linaro.org>

On Mon, 4 Sep 2023 20:36:02 +0200
Philippe Mathieu-Daudé <philmd@linaro.org> wrote:

> Hi Jonathan,
> 
> Few style comments inlined.
> 
> On 4/9/23 18:47, Jonathan Cameron wrote:
> > Support these decoders in CXL host bridges (pxb-cxl), CXL Switch USP
> > and CXL Type 3 end points.
> > 
> > Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
> > ---
Hi Philippe,

Thanks for the particularly quick reviews! 

...

> > diff --git a/hw/cxl/cxl-component-utils.c b/hw/cxl/cxl-component-utils.c
> > index e96398e8af..79b9369756 100644
> > --- a/hw/cxl/cxl-component-utils.c
> > +++ b/hw/cxl/cxl-component-utils.c
> > @@ -42,6 +42,9 @@ static void dumb_hdm_handler(CXLComponentState *cxl_cstate, hwaddr offset,
> >   
> >       switch (offset) {
> >       case A_CXL_HDM_DECODER0_CTRL:
> > +    case A_CXL_HDM_DECODER1_CTRL:
> > +    case A_CXL_HDM_DECODER2_CTRL:
> > +    case A_CXL_HDM_DECODER3_CTRL:
> >           should_commit = FIELD_EX32(value, CXL_HDM_DECODER0_CTRL, COMMIT);
> >           should_uncommit = !should_commit;
> >           break;
> > @@ -81,7 +84,7 @@ static void cxl_cache_mem_write_reg(void *opaque, hwaddr offset, uint64_t value,
> >       }
> >   
> >       if (offset >= A_CXL_HDM_DECODER_CAPABILITY &&
> > -        offset <= A_CXL_HDM_DECODER0_TARGET_LIST_HI) {
> > +        offset <= A_CXL_HDM_DECODER3_TARGET_LIST_HI) {
> >           dumb_hdm_handler(cxl_cstate, offset, value);
> >       } else {
> >           cregs->cache_mem_registers[offset / sizeof(*cregs->cache_mem_registers)] = value;
> > @@ -161,7 +164,7 @@ static void ras_init_common(uint32_t *reg_state, uint32_t *write_msk)
> >   static void hdm_init_common(uint32_t *reg_state, uint32_t *write_msk,
> >                               enum reg_type type)
> >   {
> > -    int decoder_count = 1;
> > +    int decoder_count = 4;  
> 
>    unsigned decoder_count = HDM_DECODER_COUNT;
> 
> >       int i;
> >   
> >       ARRAY_FIELD_DP32(reg_state, CXL_HDM_DECODER_CAPABILITY, DECODER_COUNT,
> > @@ -174,19 +177,22 @@ static void hdm_init_common(uint32_t *reg_state, uint32_t *write_msk,
> >                        HDM_DECODER_ENABLE, 0);
> >       write_msk[R_CXL_HDM_DECODER_GLOBAL_CONTROL] = 0x3;
> >       for (i = 0; i < decoder_count; i++) {  
> 
> Alternatively:
> 
>          for (i = 0; i < decoder_count; i++, write_msk += 8) {
>              write_msk[R_CXL_HDM_DECODER0_BASE_LO] = 0xf0000000;

That's a bit nasty and fragile given we are offsetting the base register than
indexing into it (so applying a later offset).

> 
> > -        write_msk[R_CXL_HDM_DECODER0_BASE_LO + i * 0x20] = 0xf0000000;
> > -        write_msk[R_CXL_HDM_DECODER0_BASE_HI + i * 0x20] = 0xffffffff;
> > -        write_msk[R_CXL_HDM_DECODER0_SIZE_LO + i * 0x20] = 0xf0000000;
> > -        write_msk[R_CXL_HDM_DECODER0_SIZE_HI + i * 0x20] = 0xffffffff;
> > -        write_msk[R_CXL_HDM_DECODER0_CTRL + i * 0x20] = 0x13ff;
> > +        write_msk[R_CXL_HDM_DECODER0_BASE_LO + i * 0x20 / 4] = 0xf0000000;  
> 
> (this 0x20 / 4 bugs me a bit).

Instead, I've gone with a local variable which leaves me room for deriving
this based on the step between the registers for decoders 0 and 1.

hdm_inc = R_CXL_HDM_DECODER1_BASE_LO - R_CXL_HDM_DECODER0_BASE_LO;

I haven't added a define for this because it would probably have to be
long enough that it will cause line length problems :(
So it is replicated in a few different places which isn't ideal
but definitely better than the 0x20 / 4

> 
> > +        write_msk[R_CXL_HDM_DECODER0_BASE_HI + i * 0x20 / 4]  = 0xffffffff;
> > +        write_msk[R_CXL_HDM_DECODER0_SIZE_LO + i * 0x20 / 4] = 0xf0000000;
> > +        write_msk[R_CXL_HDM_DECODER0_SIZE_HI + i * 0x20 / 4] = 0xffffffff;
> > +        write_msk[R_CXL_HDM_DECODER0_CTRL + i * 0x20 / 4] = 0x13ff;
> >           if (type == CXL2_DEVICE ||
> >               type == CXL2_TYPE3_DEVICE ||
> >               type == CXL2_LOGICAL_DEVICE) {
> > -            write_msk[R_CXL_HDM_DECODER0_TARGET_LIST_LO + i * 0x20] = 0xf0000000;
> > +            write_msk[R_CXL_HDM_DECODER0_TARGET_LIST_LO + i * 0x20 / 4] =
> > +                0xf0000000;
> >           } else {
> > -            write_msk[R_CXL_HDM_DECODER0_TARGET_LIST_LO + i * 0x20] = 0xffffffff;
> > +            write_msk[R_CXL_HDM_DECODER0_TARGET_LIST_LO + i * 0x20 / 4] =
> > +                0xffffffff;
> >           }
> > -        write_msk[R_CXL_HDM_DECODER0_TARGET_LIST_HI + i * 0x20] = 0xffffffff;
> > +        write_msk[R_CXL_HDM_DECODER0_TARGET_LIST_HI + i * 0x20 / 4] =
> > +            0xffffffff;
> >       }
> >   }  
> 



  reply	other threads:[~2023-09-05 16:24 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-09-04 16:47 [PATCH 0/2] hw/cxl: Support emulating 4 HDM decoders throughout topology Jonathan Cameron
2023-09-04 16:47 ` Jonathan Cameron via
2023-09-04 16:47 ` [PATCH 1/2] hw/cxl: Add utility functions decoder interleave ways and target count Jonathan Cameron
2023-09-04 16:47   ` Jonathan Cameron via
2023-09-04 18:26   ` Philippe Mathieu-Daudé
2023-09-05 14:56     ` Jonathan Cameron
2023-09-05 14:56       ` Jonathan Cameron via
2023-09-05 15:06       ` Jonathan Cameron
2023-09-05 15:06         ` Jonathan Cameron via
2023-09-05 16:55         ` Philippe Mathieu-Daudé
2023-09-07 11:27           ` Jonathan Cameron
2023-09-07 11:27             ` Jonathan Cameron via
2023-09-04 16:47 ` [PATCH 2/2] hw/cxl: Support 4 HDM decoders at all levels of topology Jonathan Cameron
2023-09-04 16:47   ` Jonathan Cameron via
2023-09-04 18:36   ` Philippe Mathieu-Daudé
2023-09-05 15:44     ` Jonathan Cameron [this message]
2023-09-05 15:44       ` Jonathan Cameron via

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=20230905164413.0000406c@huawei.com \
    --to=jonathan.cameron@huawei.com \
    --cc=dave.jiang@intel.com \
    --cc=fan.ni@samsung.com \
    --cc=linux-cxl@vger.kernel.org \
    --cc=linuxarm@huawei.com \
    --cc=mst@redhat.com \
    --cc=philmd@linaro.org \
    --cc=qemu-devel@nongnu.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.