Re: [Qemu-devel] [PATCH v2 42/47] acpi: add acpi_def_block() term

[Igor Mammedov]

On Thu, 29 Jan 2015 17:01:02 +0800
Shannon Zhao <address@hidden> wrote:

> On 2015/1/29 16:45, Igor Mammedov wrote:
> > On Thu, 29 Jan 2015 16:02:47 +0800
> > Shannon Zhao <address@hidden> wrote:
> > 
> >> > On 2015/1/22 22:50, Igor Mammedov wrote:
> >>> > > Signed-off-by: Igor Mammedov <address@hidden>
> >>> > > ---
> >>> > >  hw/acpi/acpi-build-utils.c         | 47 
> >>> > > ++++++++++++++++++++++++++++++++++++++
> >>> > >  hw/i386/acpi-build.c               |  1 -
> >>> > >  include/hw/acpi/acpi-build-utils.h |  8 +++++++
> >>> > >  3 files changed, 55 insertions(+), 1 deletion(-)
> >>> > > 
> >>> > > diff --git a/hw/acpi/acpi-build-utils.c b/hw/acpi/acpi-build-utils.c
> >>> > > index b19d370..58f88cd 100644
> >>> > > --- a/hw/acpi/acpi-build-utils.c
> >>> > > +++ b/hw/acpi/acpi-build-utils.c
> >>> > > @@ -26,6 +26,7 @@
> >>> > >  #include <string.h>
> >>> > >  #include "hw/acpi/acpi-build-utils.h"
> >>> > >  #include "qemu/bswap.h"
> >>> > > +#include "hw/acpi/bios-linker-loader.h"
> >>> > >  
> >>> > >  GArray *build_alloc_array(void)
> >>> > >  {
> >>> > > @@ -312,6 +313,21 @@ void aml_append(AcpiAml *parent_ctx, AcpiAml 
> >>> > > child)
> >>> > >          build_prepend_int(child.buf, child.buf->len);
> >>> > >          build_package(child.buf, child.op);
> >>> > >          break;
> >>> > > +    case DEF_BLOCK: {
> >>> > > +        uint8_t *start = (uint8_t *)parent_ctx->buf->data +
> >>> > > +                         parent_ctx->buf->len;
> >>> > > +        uint32_t le32_len = cpu_to_le32(child.buf->len);
> >>> > > +
> >>> > > +        /* create linker entry for the DefinitionBlock */
> >>> > > +        bios_linker_loader_add_checksum(parent_ctx->linker,
> >>> > > +            ACPI_BUILD_TABLE_FILE,
> >>> > > +            parent_ctx->buf->data,
> >>> > > +            start, child.buf->len, start + 9 /* checksum offset */);
> >>> > > +
> >>> > > +        /* set DefinitionBlock length at TableLength offset*/
> >>> > > +        memcpy(child.buf->data + 4, &le32_len, sizeof le32_len);
> >>> > > +        break;
> >>> > > +    }
> >>> > >      default:
> >>> > >          break;
> >>> > >      }
> >>> > > @@ -843,3 +859,34 @@ AcpiAml acpi_qword_memory(acpiDecode dec, 
> >>> > > acpiMinFixed min_fixed,
> >>> > >                                dec, addr_gran, addr_min, addr_max,
> >>> > >                                addr_trans, len, flags);
> >>> > >  }
> >>> > > +
> >>> > > +/* ACPI 5.0: 20.2.1 Table and Table Header Encoding */
> >>> > > +AcpiAml ()(const char *signature, uint8_t revision,
> >>> > > +                       const char *oem_id, const char *oem_table_id,
> >>> > > +                       uint32_t oem_revision)
> >>> > > +{
> >>> > > +    int len;
> >>> > > +    AcpiAml var = aml_allocate_internal(0, DEF_BLOCK);
> >>> > > +
> >>> > > +    assert(strlen(signature) == 4);
> >>> > > +    g_array_append_vals(var.buf, signature, 4);
> >>> > > +    build_append_value(var.buf, 0, 4); /* Length place holder */
> >>> > > +    build_append_byte(var.buf, revision);
> >>> > > +    build_append_byte(var.buf, 0); /* place holder for Checksum */
> >>> > > +
> >>> > > +    len = strlen(oem_id);
> >>> > > +    assert(len <= 6);
> >>> > > +    g_array_append_vals(var.buf, oem_id, len);
> >>> > > +    g_array_append_vals(var.buf, "\0\0\0\0\0\0\0\0", 6 - len);
> >>> > > +
> >>> > > +    len = strlen(oem_table_id);
> >>> > > +    assert(len <= 8);
> >>> > > +    g_array_append_vals(var.buf, oem_table_id, len);
> >>> > > +    g_array_append_vals(var.buf, "\0\0\0\0\0\0\0\0", 8 - len);
> >>> > > +
> >>> > > +    build_append_value(var.buf, oem_revision, 4);
> >>> > > +    g_array_append_vals(var.buf, ACPI_BUILD_APPNAME4, 4); /* 
> >>> > > asl_compiler_id */
> >>> > > +    build_append_value(var.buf, 1, 4); /* asl_compiler_revision */
> >>> > > +
> >>> > > +    return var;
> >>> > > +}
> >> > 
> >> > This function is similar with build_header() in hw/i386/acpi-build.c
> >> > But the format of omt_id, oem_table_id, asl_compiler_id are not exactly 
> >> > same.
> >> > Maybe we should make it consistent.
> > I've tried to follow APCI spec for definition block here,
> > which should be identical to build_header().
> > What exactly is not the same?
> > 
> 
> Sorry, I don make myself clear.
> I mean that the values of omt_id, oem_table_id, asl_compiler_id are different 
> between
> using build_header and acpi_def_block.
> E.g. using acpi_def_block the asl_compiler_id would be "BXPC" while 
> build_header is blank.
That's no true, build_header() sets asl_compiler_id to ACPI_BUILD_APPNAME4
the same way as does acpi_def_block().

> 
> I think we should make the omt_id, oem_table_id, asl_compiler_id as 
> parameters of the
> functions, then users assign the value.
The only thing not exposed to API user is asl_compiler_id and
asl_compiler_revision, I'll add them as arguments of acpi_def_block()
on the next respin to be closer to ACPI spec.

> 
> Thanks,
> Shannon
>