qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Jonathan Cameron via <qemu-devel@nongnu.org>
To: "Michael S. Tsirkin" <mst@redhat.com>
Cc: qemu-devel@nongnu.org, linux-cxl@vger.kernel.org,
	"Michael Tokarev" <mjt@tls.msk.ru>,
	linuxarm@huawei.com, "Fan Ni" <fan.ni@samsung.com>,
	"Philippe Mathieu-Daudé" <philmd@linaro.org>
Subject: Re: [PATCH v4 2/4] hw/cxl: Use switch statements for read and write of cachemem registers
Date: Wed, 18 Oct 2023 17:31:42 +0100	[thread overview]
Message-ID: <20231018173142.00003739@Huawei.com> (raw)
In-Reply-To: <20231018172239.00005f3d@Huawei.com>

On Wed, 18 Oct 2023 17:22:39 +0100
Jonathan Cameron <Jonathan.Cameron@Huawei.com> wrote:

> On Wed, 18 Oct 2023 10:40:45 -0400
> "Michael S. Tsirkin" <mst@redhat.com> wrote:
> 
> > On Thu, Oct 12, 2023 at 03:05:12PM +0100, Jonathan Cameron wrote:  
> > > Establishing that only register accesses of size 4 and 8 can occur
> > > using these functions requires looking at their callers. Make it
> > > easier to see that by using switch statements.
> > > Assertions are used to enforce that the register storage is of the
> > > matching size, allowing fixed values to be used for divisors of
> > > the array indices.
> > > 
> > > Suggested-by: Michael Tokarev <mjt@tls.msk.ru>
> > > Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
> > > Reviewed-by: Fan Ni <fan.ni@samsung.com>    
> > 
> > Fails with clang:
> > 
> > ../hw/cxl/cxl-component-utils.c:130:9: error: expected expression
> >         QEMU_BUILD_BUG_ON(sizeof(*cregs->cache_mem_regs_write_mask) != 4);
> >         ^
> > ../include/qemu/compiler.h:76:30: note: expanded from macro 'QEMU_BUILD_BUG_ON'
> > #define QEMU_BUILD_BUG_ON(x) QEMU_BUILD_BUG_MSG(x, "not expecting: " #x)
> >                              ^
> > ../include/qemu/compiler.h:74:36: note: expanded from macro 'QEMU_BUILD_BUG_MSG'
> > #define QEMU_BUILD_BUG_MSG(x, msg) _Static_assert(!(x), msg)
> >                                    ^
> > 1 error generated.
> > 
> > See e.g.
> > 
> > https://gitlab.com/mstredhat/qemu/-/jobs/5320879531  
> 
> Thanks. Will see if I can figure out what is going wrong.
> This code seems cursed.  I need to figure out a way to run the CI again.
> Ran into credit problems I couldn't solve last time I tried.
> 
> In meantime I can replicate this locally but not immediately obvious to me
> why it is failing.

Any clang experts around?  This seems to be fixed by reducing the scope

case 4:
{
	QEMU_BUILD_BUG_ON();
}
I have no idea why though...


> 
> Jonathan
> 
> > 
> > 
> >   
> > > ---
> > > v4: Use QEMU_BUILD_BUG_ON() instead of static_assert with missing
> > >     error message.
> > > 
> > >  hw/cxl/cxl-component-utils.c | 65 +++++++++++++++++++++++-------------
> > >  1 file changed, 42 insertions(+), 23 deletions(-)
> > > 
> > > diff --git a/hw/cxl/cxl-component-utils.c b/hw/cxl/cxl-component-utils.c
> > > index f3bbf0fd13..d1997a52e6 100644
> > > --- a/hw/cxl/cxl-component-utils.c
> > > +++ b/hw/cxl/cxl-component-utils.c
> > > @@ -67,16 +67,24 @@ static uint64_t cxl_cache_mem_read_reg(void *opaque, hwaddr offset,
> > >      CXLComponentState *cxl_cstate = opaque;
> > >      ComponentRegisters *cregs = &cxl_cstate->crb;
> > >  
> > > -    if (size == 8) {
> > > +    switch (size) {
> > > +    case 4:
> > > +        if (cregs->special_ops && cregs->special_ops->read) {
> > > +            return cregs->special_ops->read(cxl_cstate, offset, 4);
> > > +        } else {
> > > +            QEMU_BUILD_BUG_ON(sizeof(*cregs->cache_mem_registers) != 4);
> > > +            return cregs->cache_mem_registers[offset / 4];
> > > +        }
> > > +    case 8:
> > >          qemu_log_mask(LOG_UNIMP,
> > >                        "CXL 8 byte cache mem registers not implemented\n");
> > >          return 0;
> > > -    }
> > > -
> > > -    if (cregs->special_ops && cregs->special_ops->read) {
> > > -        return cregs->special_ops->read(cxl_cstate, offset, size);
> > > -    } else {
> > > -        return cregs->cache_mem_registers[offset / sizeof(*cregs->cache_mem_registers)];
> > > +    default:
> > > +        /*
> > > +         * In line with specifiction limitaions on access sizes, this
> > > +         * routine is not called with other sizes.
> > > +         */
> > > +        g_assert_not_reached();
> > >      }
> > >  }
> > >  
> > > @@ -117,25 +125,36 @@ static void cxl_cache_mem_write_reg(void *opaque, hwaddr offset, uint64_t value,
> > >      ComponentRegisters *cregs = &cxl_cstate->crb;
> > >      uint32_t mask;
> > >  
> > > -    if (size == 8) {
> > > +    switch (size) {
> > > +    case 4:
> > > +        QEMU_BUILD_BUG_ON(sizeof(*cregs->cache_mem_regs_write_mask) != 4);
> > > +        QEMU_BUILD_BUG_ON(sizeof(*cregs->cache_mem_registers) != 4);
> > > +        mask = cregs->cache_mem_regs_write_mask[offset / 4];
> > > +        value &= mask;
> > > +        /* RO bits should remain constant. Done by reading existing value */
> > > +        value |= ~mask & cregs->cache_mem_registers[offset / 4];
> > > +        if (cregs->special_ops && cregs->special_ops->write) {
> > > +            cregs->special_ops->write(cxl_cstate, offset, value, size);
> > > +            return;
> > > +        }
> > > +
> > > +        if (offset >= A_CXL_HDM_DECODER_CAPABILITY &&
> > > +            offset <= A_CXL_HDM_DECODER3_TARGET_LIST_HI) {
> > > +            dumb_hdm_handler(cxl_cstate, offset, value);
> > > +        } else {
> > > +            cregs->cache_mem_registers[offset / 4] = value;
> > > +        }
> > > +        return;
> > > +    case 8:
> > >          qemu_log_mask(LOG_UNIMP,
> > >                        "CXL 8 byte cache mem registers not implemented\n");
> > >          return;
> > > -    }
> > > -    mask = cregs->cache_mem_regs_write_mask[offset / sizeof(*cregs->cache_mem_regs_write_mask)];
> > > -    value &= mask;
> > > -    /* RO bits should remain constant. Done by reading existing value */
> > > -    value |= ~mask & cregs->cache_mem_registers[offset / sizeof(*cregs->cache_mem_registers)];
> > > -    if (cregs->special_ops && cregs->special_ops->write) {
> > > -        cregs->special_ops->write(cxl_cstate, offset, value, size);
> > > -        return;
> > > -    }
> > > -
> > > -    if (offset >= A_CXL_HDM_DECODER_CAPABILITY &&
> > > -        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;
> > > +    default:
> > > +        /*
> > > +         * In line with specifiction limitaions on access sizes, this
> > > +         * routine is not called with other sizes.
> > > +         */
> > > +        g_assert_not_reached();
> > >      }
> > >  }
> > >  
> > > -- 
> > > 2.39.2    
> > 
> >   
> 



WARNING: multiple messages have this Message-ID (diff)
From: Jonathan Cameron <Jonathan.Cameron@Huawei.com>
To: "Michael S. Tsirkin" <mst@redhat.com>
Cc: qemu-devel@nongnu.org, linux-cxl@vger.kernel.org,
	"Michael Tokarev" <mjt@tls.msk.ru>,
	linuxarm@huawei.com, "Fan Ni" <fan.ni@samsung.com>,
	"Philippe Mathieu-Daudé" <philmd@linaro.org>
Subject: Re: [PATCH v4 2/4] hw/cxl: Use switch statements for read and write of cachemem registers
Date: Wed, 18 Oct 2023 17:31:42 +0100	[thread overview]
Message-ID: <20231018173142.00003739@Huawei.com> (raw)
Message-ID: <20231018163142.O18Lj5Iu3GMnLJ0UGZ8McpsDjeYpvjPHuoRHct-oRN0@z> (raw)
In-Reply-To: <20231018172239.00005f3d@Huawei.com>

On Wed, 18 Oct 2023 17:22:39 +0100
Jonathan Cameron <Jonathan.Cameron@Huawei.com> wrote:

> On Wed, 18 Oct 2023 10:40:45 -0400
> "Michael S. Tsirkin" <mst@redhat.com> wrote:
> 
> > On Thu, Oct 12, 2023 at 03:05:12PM +0100, Jonathan Cameron wrote:  
> > > Establishing that only register accesses of size 4 and 8 can occur
> > > using these functions requires looking at their callers. Make it
> > > easier to see that by using switch statements.
> > > Assertions are used to enforce that the register storage is of the
> > > matching size, allowing fixed values to be used for divisors of
> > > the array indices.
> > > 
> > > Suggested-by: Michael Tokarev <mjt@tls.msk.ru>
> > > Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
> > > Reviewed-by: Fan Ni <fan.ni@samsung.com>    
> > 
> > Fails with clang:
> > 
> > ../hw/cxl/cxl-component-utils.c:130:9: error: expected expression
> >         QEMU_BUILD_BUG_ON(sizeof(*cregs->cache_mem_regs_write_mask) != 4);
> >         ^
> > ../include/qemu/compiler.h:76:30: note: expanded from macro 'QEMU_BUILD_BUG_ON'
> > #define QEMU_BUILD_BUG_ON(x) QEMU_BUILD_BUG_MSG(x, "not expecting: " #x)
> >                              ^
> > ../include/qemu/compiler.h:74:36: note: expanded from macro 'QEMU_BUILD_BUG_MSG'
> > #define QEMU_BUILD_BUG_MSG(x, msg) _Static_assert(!(x), msg)
> >                                    ^
> > 1 error generated.
> > 
> > See e.g.
> > 
> > https://gitlab.com/mstredhat/qemu/-/jobs/5320879531  
> 
> Thanks. Will see if I can figure out what is going wrong.
> This code seems cursed.  I need to figure out a way to run the CI again.
> Ran into credit problems I couldn't solve last time I tried.
> 
> In meantime I can replicate this locally but not immediately obvious to me
> why it is failing.

Any clang experts around?  This seems to be fixed by reducing the scope

case 4:
{
	QEMU_BUILD_BUG_ON();
}
I have no idea why though...


> 
> Jonathan
> 
> > 
> > 
> >   
> > > ---
> > > v4: Use QEMU_BUILD_BUG_ON() instead of static_assert with missing
> > >     error message.
> > > 
> > >  hw/cxl/cxl-component-utils.c | 65 +++++++++++++++++++++++-------------
> > >  1 file changed, 42 insertions(+), 23 deletions(-)
> > > 
> > > diff --git a/hw/cxl/cxl-component-utils.c b/hw/cxl/cxl-component-utils.c
> > > index f3bbf0fd13..d1997a52e6 100644
> > > --- a/hw/cxl/cxl-component-utils.c
> > > +++ b/hw/cxl/cxl-component-utils.c
> > > @@ -67,16 +67,24 @@ static uint64_t cxl_cache_mem_read_reg(void *opaque, hwaddr offset,
> > >      CXLComponentState *cxl_cstate = opaque;
> > >      ComponentRegisters *cregs = &cxl_cstate->crb;
> > >  
> > > -    if (size == 8) {
> > > +    switch (size) {
> > > +    case 4:
> > > +        if (cregs->special_ops && cregs->special_ops->read) {
> > > +            return cregs->special_ops->read(cxl_cstate, offset, 4);
> > > +        } else {
> > > +            QEMU_BUILD_BUG_ON(sizeof(*cregs->cache_mem_registers) != 4);
> > > +            return cregs->cache_mem_registers[offset / 4];
> > > +        }
> > > +    case 8:
> > >          qemu_log_mask(LOG_UNIMP,
> > >                        "CXL 8 byte cache mem registers not implemented\n");
> > >          return 0;
> > > -    }
> > > -
> > > -    if (cregs->special_ops && cregs->special_ops->read) {
> > > -        return cregs->special_ops->read(cxl_cstate, offset, size);
> > > -    } else {
> > > -        return cregs->cache_mem_registers[offset / sizeof(*cregs->cache_mem_registers)];
> > > +    default:
> > > +        /*
> > > +         * In line with specifiction limitaions on access sizes, this
> > > +         * routine is not called with other sizes.
> > > +         */
> > > +        g_assert_not_reached();
> > >      }
> > >  }
> > >  
> > > @@ -117,25 +125,36 @@ static void cxl_cache_mem_write_reg(void *opaque, hwaddr offset, uint64_t value,
> > >      ComponentRegisters *cregs = &cxl_cstate->crb;
> > >      uint32_t mask;
> > >  
> > > -    if (size == 8) {
> > > +    switch (size) {
> > > +    case 4:
> > > +        QEMU_BUILD_BUG_ON(sizeof(*cregs->cache_mem_regs_write_mask) != 4);
> > > +        QEMU_BUILD_BUG_ON(sizeof(*cregs->cache_mem_registers) != 4);
> > > +        mask = cregs->cache_mem_regs_write_mask[offset / 4];
> > > +        value &= mask;
> > > +        /* RO bits should remain constant. Done by reading existing value */
> > > +        value |= ~mask & cregs->cache_mem_registers[offset / 4];
> > > +        if (cregs->special_ops && cregs->special_ops->write) {
> > > +            cregs->special_ops->write(cxl_cstate, offset, value, size);
> > > +            return;
> > > +        }
> > > +
> > > +        if (offset >= A_CXL_HDM_DECODER_CAPABILITY &&
> > > +            offset <= A_CXL_HDM_DECODER3_TARGET_LIST_HI) {
> > > +            dumb_hdm_handler(cxl_cstate, offset, value);
> > > +        } else {
> > > +            cregs->cache_mem_registers[offset / 4] = value;
> > > +        }
> > > +        return;
> > > +    case 8:
> > >          qemu_log_mask(LOG_UNIMP,
> > >                        "CXL 8 byte cache mem registers not implemented\n");
> > >          return;
> > > -    }
> > > -    mask = cregs->cache_mem_regs_write_mask[offset / sizeof(*cregs->cache_mem_regs_write_mask)];
> > > -    value &= mask;
> > > -    /* RO bits should remain constant. Done by reading existing value */
> > > -    value |= ~mask & cregs->cache_mem_registers[offset / sizeof(*cregs->cache_mem_registers)];
> > > -    if (cregs->special_ops && cregs->special_ops->write) {
> > > -        cregs->special_ops->write(cxl_cstate, offset, value, size);
> > > -        return;
> > > -    }
> > > -
> > > -    if (offset >= A_CXL_HDM_DECODER_CAPABILITY &&
> > > -        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;
> > > +    default:
> > > +        /*
> > > +         * In line with specifiction limitaions on access sizes, this
> > > +         * routine is not called with other sizes.
> > > +         */
> > > +        g_assert_not_reached();
> > >      }
> > >  }
> > >  
> > > -- 
> > > 2.39.2    
> > 
> >   
> 



  parent reply	other threads:[~2023-10-18 16:32 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-10-12 14:05 [PATCH v4 0/4] hw/cxl: Line length reduction and related Jonathan Cameron via
2023-10-12 14:05 ` Jonathan Cameron
2023-10-12 14:05 ` [PATCH v4 1/4] hw/cxl: Use a switch to explicitly check size in caps_reg_read() Jonathan Cameron via
2023-10-12 14:05   ` Jonathan Cameron
2023-10-12 14:05 ` [PATCH v4 2/4] hw/cxl: Use switch statements for read and write of cachemem registers Jonathan Cameron via
2023-10-12 14:05   ` Jonathan Cameron
2023-10-18 14:40   ` Michael S. Tsirkin
2023-10-18 16:22     ` Jonathan Cameron via
2023-10-18 16:22       ` Jonathan Cameron
2023-10-18 16:31       ` Jonathan Cameron via [this message]
2023-10-18 16:31         ` Jonathan Cameron
2023-10-19  9:09         ` Jonathan Cameron via
2023-10-19  9:09           ` Jonathan Cameron
2023-10-19  9:23           ` Michael S. Tsirkin
2023-10-19 11:24             ` Jonathan Cameron via
2023-10-19 11:24               ` Jonathan Cameron
2023-10-19 11:48               ` Jonathan Cameron via
2023-10-19 11:48                 ` Jonathan Cameron
2023-10-12 14:05 ` [PATCH v4 3/4] hw/cxl: CXLDVSECPortExtensions renamed to CXLDVSECPortExt Jonathan Cameron via
2023-10-12 14:05   ` Jonathan Cameron
2023-10-12 14:05 ` [PATCH v4 4/4] hw/cxl: Line length reductions Jonathan Cameron via
2023-10-12 14:05   ` Jonathan Cameron
2023-10-13 16:56   ` fan

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=20231018173142.00003739@Huawei.com \
    --to=qemu-devel@nongnu.org \
    --cc=Jonathan.Cameron@Huawei.com \
    --cc=fan.ni@samsung.com \
    --cc=linux-cxl@vger.kernel.org \
    --cc=linuxarm@huawei.com \
    --cc=mjt@tls.msk.ru \
    --cc=mst@redhat.com \
    --cc=philmd@linaro.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).