Skip to contentSkip to conversation

xen/acpi: Don't fail if SPCR table is absent

PATCH In review Elliott Mitchell 4 participants 10 messages Oct 21, 2020 – Oct 23, 2020

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 xen/acpi: Don't fail if SPCR table is absent

Absence of a SPCR table likely means the console is a framebuffer. In such case acpi_iomem_deny_access() should NOT fail.

---
 xen/arch/arm/acpi/domain_build.c | 19 ++++++++++---------
 1 file changed, 10 insertions(+), 9 deletions(-)
xen/arch/arm/acpi/domain_build.c +10 -9
index 1b1cfabb00..bbdc90f92c 100644
@@ -42,17 +42,18 @@ static int __init acpi_iomem_deny_access(struct domain *d)
     status = acpi_get_table(ACPI_SIG_SPCR, 0,
                             (struct acpi_table_header **)&spcr);
 
-    if ( ACPI_FAILURE(status) )
+    if ( ACPI_SUCCESS(status) )
     {
-        printk("Failed to get SPCR table\n");
-        return -EINVAL;
+        mfn = spcr->serial_port.address >> PAGE_SHIFT;
+        /* Deny MMIO access for UART */
+        rc = iomem_deny_access(d, mfn, mfn + 1);
+        if ( rc )
+            return rc;
+    }
+    else
+    {
+        printk("Failed to get SPCR table, Xen console may be unavailable\n");
     }
-
-    mfn = spcr->serial_port.address >> PAGE_SHIFT;
-    /* Deny MMIO access for UART */
-    rc = iomem_deny_access(d, mfn, mfn + 1);
-    if ( rc )
-        return rc;
 
     /* Deny MMIO access for GIC regions */
     return gic_iomem_deny_access(d);
Signature · 9 lines
2.20.1



-- 
(\___(\___(\______          --=> 8-) EHM <=--          ______/)___/)___/)
 \BS (    |         [email protected]  PGP 87145445         |    )   /
  \_CS\   |  _____  -O #include <stddisclaimer.h> O-   _____  |   /  _/
8A19\___\_|_/58D2 7E3D DDF4 7BA6 <-PGP-> 41D1 B375 37D0 8714\_|_/___/5445
Reviewed-by on v1 1/1
#1 lore raw
Quoted 4 lines · Elliott Mitchell

Absence of a SPCR table likely means the console is a framebuffer. In such case acpi_iomem_deny_access() should NOT fail.

Quoted 36 lines
---
 xen/arch/arm/acpi/domain_build.c | 19 ++++++++++---------
 1 file changed, 10 insertions(+), 9 deletions(-)
xen/arch/arm/acpi/domain_build.c +10 -9
index 1b1cfabb00..bbdc90f92c 100644
@@ -42,17 +42,18 @@ static int __init acpi_iomem_deny_access(struct domain *d)
     status = acpi_get_table(ACPI_SIG_SPCR, 0,
                             (struct acpi_table_header **)&spcr);
 
-    if ( ACPI_FAILURE(status) )
+    if ( ACPI_SUCCESS(status) )
     {
-        printk("Failed to get SPCR table\n");
-        return -EINVAL;
+        mfn = spcr->serial_port.address >> PAGE_SHIFT;
+        /* Deny MMIO access for UART */
+        rc = iomem_deny_access(d, mfn, mfn + 1);
+        if ( rc )
+            return rc;
+    }
+    else
+    {
+        printk("Failed to get SPCR table, Xen console may be unavailable\n");
     }
-
-    mfn = spcr->serial_port.address >> PAGE_SHIFT;
-    /* Deny MMIO access for UART */
-    rc = iomem_deny_access(d, mfn, mfn + 1);
-    if ( rc )
-        return rc;
 
     /* Deny MMIO access for GIC regions */
     return gic_iomem_deny_access(d);
#2 lore raw
In reply to Elliott Mitchell
Quoted 11 lines · Elliott Mitchell
xen/arch/arm/acpi/domain_build.c +1 -3
@@ -42,17 +42,18 @@ static int __init acpi_iomem_deny_access(struct domain *d)
     status = acpi_get_table(ACPI_SIG_SPCR, 0,
                             (struct acpi_table_header **)&spcr);
 
-    if ( ACPI_FAILURE(status) )
+    if ( ACPI_SUCCESS(status) )
     {
-        printk("Failed to get SPCR table\n");
-        return -EINVAL;

xen/arch/arm/acpi/domain_build.c static int __init acpi_iomem_deny_access(struct domain *d) · line 56

Quoted 10 lines
xen/arch/arm/acpi/domain_build.c +9 -0
+        mfn = spcr->serial_port.address >> PAGE_SHIFT;
+        /* Deny MMIO access for UART */
+        rc = iomem_deny_access(d, mfn, mfn + 1);
+        if ( rc )
+            return rc;
+    }
+    else
+    {
+        printk("Failed to get SPCR table, Xen console may be unavailable\n");
     }

Nit: While I see you've got Stefano's R-b already, I Xen we typically omit the braces here.

Jan

#3 lore raw
Quoted 17 lines · Jan Beulich
Quoted 16 lines · Elliott Mitchell
xen/arch/arm/acpi/domain_build.c +6 -3
@@ -42,17 +42,18 @@ static int __init acpi_iomem_deny_access(struct domain *d)
     status = acpi_get_table(ACPI_SIG_SPCR, 0,
                             (struct acpi_table_header **)&spcr);
 
-    if ( ACPI_FAILURE(status) )
+    if ( ACPI_SUCCESS(status) )
     {
-        printk("Failed to get SPCR table\n");
-        return -EINVAL;
+        mfn = spcr->serial_port.address >> PAGE_SHIFT;
+        /* Deny MMIO access for UART */
+        rc = iomem_deny_access(d, mfn, mfn + 1);
+        if ( rc )
+            return rc;
Quoted 8 lines

xen/arch/arm/acpi/domain_build.c static int __init acpi_iomem_deny_access(struct domain *d) · line 56

Quoted 5 lines
xen/arch/arm/acpi/domain_build.c +4 -0
+    }
+    else
+    {
+        printk("Failed to get SPCR table, Xen console may be unavailable\n");
     }

Nit: While I see you've got Stefano's R-b already, I Xen we typically omit the braces here.

Personally, I prefer that myself, but was unsure of the preference here. I've seen multiple projects which *really* dislike using having brackets for some clauses, but not others (ie they want either all clauses with or all clauses without; instead of only if required).

I sent what I thought was the more often used format. (I also like tabs, and dislike having so many spaces; alas my preferences are apparently uncommon)

Signature · 4 lines
(\___(\___(\______          --=> 8-) EHM <=--          ______/)___/)___/)
 \BS (    |         [email protected]  PGP 87145445         |    )   /
  \_CS\   |  _____  -O #include <stddisclaimer.h> O-   _____  |   /  _/
8A19\___\_|_/58D2 7E3D DDF4 7BA6 <-PGP-> 41D1 B375 37D0 8714\_|_/___/5445
Acked-by
#4 lore raw

Hi Elliott,

Quoted 27 lines · Elliott Mitchell
Quoted 17 lines · Jan Beulich
Quoted 16 lines · Elliott Mitchell
xen/arch/arm/acpi/domain_build.c +6 -3
@@ -42,17 +42,18 @@ static int __init acpi_iomem_deny_access(struct domain *d)
      status = acpi_get_table(ACPI_SIG_SPCR, 0,
                              (struct acpi_table_header **)&spcr);
 
-    if ( ACPI_FAILURE(status) )
+    if ( ACPI_SUCCESS(status) )
      {
-        printk("Failed to get SPCR table\n");
-        return -EINVAL;
+        mfn = spcr->serial_port.address >> PAGE_SHIFT;
+        /* Deny MMIO access for UART */
+        rc = iomem_deny_access(d, mfn, mfn + 1);
+        if ( rc )
+            return rc;
Quoted 8 lines

xen/arch/arm/acpi/domain_build.c static int __init acpi_iomem_deny_access(struct domain *d) · line 56

Quoted 5 lines
xen/arch/arm/acpi/domain_build.c +4 -0
+    }
+    else
+    {
+        printk("Failed to get SPCR table, Xen console may be unavailable\n");
      }

Nit: While I see you've got Stefano's R-b already, I Xen we typically omit the braces here.

Quoted 1 line

Personally, I prefer that myself, but was unsure of the preference here.

I don't think we are very consistent here... I would not add them myself, but I don't particularly mind them (I know some editors will add them automatically).

I will keep them while committing. For the patch:

Quoted 7 lines

I've seen multiple projects which *really* dislike using having brackets for some clauses, but not others (ie they want either all clauses with or all clauses without; instead of only if required).

I sent what I thought was the more often used format. (I also like tabs, and dislike having so many spaces; alas my preferences are apparently uncommon)

We have a few files in Xen using tabs (yes, we like mixing coding style!).

Cheers,

Signature · 1 line
Julien Grall
#5 lore raw
In reply to Elliott Mitchell
Quoted 27 lines · Elliott Mitchell
Quoted 17 lines · Jan Beulich
Quoted 16 lines · Elliott Mitchell
xen/arch/arm/acpi/domain_build.c +6 -3
@@ -42,17 +42,18 @@ static int __init acpi_iomem_deny_access(struct domain *d)
     status = acpi_get_table(ACPI_SIG_SPCR, 0,
                             (struct acpi_table_header **)&spcr);
 
-    if ( ACPI_FAILURE(status) )
+    if ( ACPI_SUCCESS(status) )
     {
-        printk("Failed to get SPCR table\n");
-        return -EINVAL;
+        mfn = spcr->serial_port.address >> PAGE_SHIFT;
+        /* Deny MMIO access for UART */
+        rc = iomem_deny_access(d, mfn, mfn + 1);
+        if ( rc )
+            return rc;
Quoted 8 lines

xen/arch/arm/acpi/domain_build.c static int __init acpi_iomem_deny_access(struct domain *d) · line 56

Quoted 5 lines
xen/arch/arm/acpi/domain_build.c +4 -0
+    }
+    else
+    {
+        printk("Failed to get SPCR table, Xen console may be unavailable\n");
     }

Nit: While I see you've got Stefano's R-b already, I Xen we typically omit the braces here.

Quoted 8 lines

Personally, I prefer that myself, but was unsure of the preference here. I've seen multiple projects which *really* dislike using having brackets for some clauses, but not others (ie they want either all clauses with or all clauses without; instead of only if required).

I sent what I thought was the more often used format. (I also like tabs, and dislike having so many spaces; alas my preferences are apparently uncommon)

"More often used" doesn't matter when there's an explicit statement about this in ./CODING_STYLE: "Braces should be omitted for blocks with a single statement." Yes, there are projects requiring all if/else-if belonging together to consistently use or not use braces. But there's explicitly no such statement in our doc. (Along these lines, dislike of spaces [and favoring tabs] also doesn't matter, as again the doc is explicit about it.)

Jan

#6 lore raw
In reply to Elliott Mitchell

Hi,

Thank you for the patch. FIY I tweak a bit the commit title before committing.

The title is now: "xen/arm: acpi: Don't fail it SPCR table is absent".

Cheers,

Quoted 41 lines · Elliott Mitchell

Absence of a SPCR table likely means the console is a framebuffer. In such case acpi_iomem_deny_access() should NOT fail.

---
  xen/arch/arm/acpi/domain_build.c | 19 ++++++++++---------
  1 file changed, 10 insertions(+), 9 deletions(-)
xen/arch/arm/acpi/domain_build.c +10 -9
index 1b1cfabb00..bbdc90f92c 100644
@@ -42,17 +42,18 @@ static int __init acpi_iomem_deny_access(struct domain *d)
      status = acpi_get_table(ACPI_SIG_SPCR, 0,
                              (struct acpi_table_header **)&spcr);
 
-    if ( ACPI_FAILURE(status) )
+    if ( ACPI_SUCCESS(status) )
      {
-        printk("Failed to get SPCR table\n");
-        return -EINVAL;
+        mfn = spcr->serial_port.address >> PAGE_SHIFT;
+        /* Deny MMIO access for UART */
+        rc = iomem_deny_access(d, mfn, mfn + 1);
+        if ( rc )
+            return rc;
+    }
+    else
+    {
+        printk("Failed to get SPCR table, Xen console may be unavailable\n");
      }
-
-    mfn = spcr->serial_port.address >> PAGE_SHIFT;
-    /* Deny MMIO access for UART */
-    rc = iomem_deny_access(d, mfn, mfn + 1);
-    if ( rc )
-        return rc;
 
      /* Deny MMIO access for GIC regions */
      return gic_iomem_deny_access(d);
Signature · 1 line
Julien Grall
#7 lore raw
Quoted 5 lines · Julien Grall

I don't think we are very consistent here... I would not add them myself, but I don't particularly mind them (I know some editors will add them automatically).

I will keep them while committing. For the patch:

I would tend to remove them on commit since I dislike them. Just as stated, I was unsure.

On default settings, clang-format will object to:

if(thing)
{
foo
}
else
bar

Or

if(thing)
foo
else
{
bar
}

I *like* those formats, but was under the impression most people did not. The indentation is the more visually obvious indicator, just the compiler actually uses the brackets. As such I *like* the misleading indentation warnings as those seemed to have a fairly high true-positive rate.

Quoted 4 lines · Julien Grall

Thank you for the patch. FIY I tweak a bit the commit title before committing.

The title is now: "xen/arm: acpi: Don't fail it SPCR table is absent".

Perhaps "xen/arm: acpi: Don't fail on absent SPCR table"?

What you're suggesting doesn't read well to me.

Signature · 4 lines
(\___(\___(\______          --=> 8-) EHM <=--          ______/)___/)___/)
 \BS (    |         [email protected]  PGP 87145445         |    )   /
  \_CS\   |  _____  -O #include <stddisclaimer.h> O-   _____  |   /  _/
8A19\___\_|_/58D2 7E3D DDF4 7BA6 <-PGP-> 41D1 B375 37D0 8714\_|_/___/5445
#8 lore raw
Quoted 9 lines · Elliott Mitchell
Quoted 4 lines · Julien Grall

Thank you for the patch. FIY I tweak a bit the commit title before committing.

The title is now: "xen/arm: acpi: Don't fail it SPCR table is absent".

Perhaps "xen/arm: acpi: Don't fail on absent SPCR table"?

What you're suggesting doesn't read well to me.

Perhaps Julien meant "if" instead of "it". i.e. a simple typo?

Jan

#9 lore raw
In reply to Elliott Mitchell

Hi Elliott,

Quoted 10 lines · Elliott Mitchell

On Thu, Oct 22, 2020 at 07:38:26PM +0100, Julien Grall wrote:

Quoted 4 lines · Julien Grall

Thank you for the patch. FIY I tweak a bit the commit title before committing.

The title is now: "xen/arm: acpi: Don't fail it SPCR table is absent".

Perhaps "xen/arm: acpi: Don't fail on absent SPCR table"?

What you're suggesting doesn't read well to me.

Sorry, I made a typo when writing the title in the e-mail. Here a direct copy from the commit:

"xen/arm: acpi: Don't fail if SPCR table is absent"

This is pretty much your original title with "arm: " added to clarify the subsystem modified.

Cheers,

Signature · 1 line
Julien Grall