[PATCH] docs and cleanups for v3.6 (v1)

[PATCH] docs and cleanups for v3.6 (v1)

[Konrad Rzeszutek Wilk]

A little of docs, cleanups of code, and some extra bits to make
reporting more pleasing to eyes.

 arch/x86/xen/enlighten.c |   41 ++++++++++++++++++++++++++++++---
 arch/x86/xen/mmu.c       |   25 ++++++++++++++------
 arch/x86/xen/p2m.c       |   19 ++++++++++-----
 arch/x86/xen/setup.c     |    9 -------
 arch/x86/xen/xen-head.S  |   56 ++++++++++++++++++++++++++++++++++++++++++++-
 arch/x86/xen/xen-ops.h   |    2 +-
 6 files changed, 121 insertions(+), 31 deletions(-)

[PATCH 1/5] xen/p2m: Fix the comment describing the P2M tree.

[Konrad Rzeszutek Wilk]

It mixed up the p2m_mid_missing with p2m_missing. Also
remove some extra spaces.

Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
---
 arch/x86/xen/p2m.c |   14 +++++++-------
 1 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/arch/x86/xen/p2m.c b/arch/x86/xen/p2m.c
index 64effdc..e4adbfb 100644
--- a/arch/x86/xen/p2m.c
+++ b/arch/x86/xen/p2m.c
@@ -22,7 +22,7 @@
  *
  * P2M_PER_PAGE depends on the architecture, as a mfn is always
  * unsigned long (8 bytes on 64-bit, 4 bytes on 32), leading to
- * 512 and 1024 entries respectively. 
+ * 512 and 1024 entries respectively.
  *
  * In short, these structures contain the Machine Frame Number (MFN) of the PFN.
  *
@@ -139,11 +139,11 @@
  *      /    | ~0, ~0, ....  |
  *     |     \---------------/
  *     |
- *     p2m_missing             p2m_missing
- * /------------------\     /------------\
- * | [p2m_mid_missing]+---->| ~0, ~0, ~0 |
- * | [p2m_mid_missing]+---->| ..., ~0    |
- * \------------------/     \------------/
+ *   p2m_mid_missing           p2m_missing
+ * /-----------------\     /------------\
+ * | [p2m_missing]   +---->| ~0, ~0, ~0 |
+ * | [p2m_missing]   +---->| ..., ~0    |
+ * \-----------------/     \------------/
  *
  * where ~0 is INVALID_P2M_ENTRY. IDENTITY is (PFN | IDENTITY_BIT)
  */
@@ -423,7 +423,7 @@ static void free_p2m_page(void *p)
 	free_page((unsigned long)p);
 }
 
-/* 
+/*
  * Fully allocate the p2m structure for a given pfn.  We need to check
  * that both the top and mid levels are allocated, and make sure the
  * parallel mfn tree is kept in sync.  We may race with other cpus, so
-- 
1.7.7.6

[PATCH 2/5] xen/mmu: The xen_setup_kernel_pagetable doesn't need to return anything.

[Konrad Rzeszutek Wilk]

We don't need to return the new PGD - as we do not use it.

Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
---
 arch/x86/xen/enlighten.c |    5 +----
 arch/x86/xen/mmu.c       |   10 ++--------
 arch/x86/xen/xen-ops.h   |    2 +-
 3 files changed, 4 insertions(+), 13 deletions(-)

diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index ed7d549..ae8a2ab 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -1256,7 +1256,6 @@ asmlinkage void __init xen_start_kernel(void)
 {
 	struct physdev_set_iopl set_iopl;
 	int rc;
-	pgd_t *pgd;
 
 	if (!xen_start_info)
 		return;
@@ -1348,8 +1347,6 @@ asmlinkage void __init xen_start_kernel(void)
 	acpi_numa = -1;
 #endif
 
-	pgd = (pgd_t *)xen_start_info->pt_base;
-
 	/* Don't do the full vcpu_info placement stuff until we have a
 	   possible map and a non-dummy shared_info. */
 	per_cpu(xen_vcpu, 0) = &HYPERVISOR_shared_info->vcpu_info[0];
@@ -1358,7 +1355,7 @@ asmlinkage void __init xen_start_kernel(void)
 	early_boot_irqs_disabled = true;
 
 	xen_raw_console_write("mapping kernel into physical memory\n");
-	pgd = xen_setup_kernel_pagetable(pgd, xen_start_info->nr_pages);
+	xen_setup_kernel_pagetable((pgd_t *)xen_start_info->pt_base, xen_start_info->nr_pages);
 
 	/* Allocate and initialize top and mid mfn levels for p2m structure */
 	xen_build_mfn_list_list();
diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c
index 3a73785..4ac21a4 100644
--- a/arch/x86/xen/mmu.c
+++ b/arch/x86/xen/mmu.c
@@ -1719,8 +1719,7 @@ static void convert_pfn_mfn(void *v)
  * of the physical mapping once some sort of allocator has been set
  * up.
  */
-pgd_t * __init xen_setup_kernel_pagetable(pgd_t *pgd,
-					 unsigned long max_pfn)
+void __init xen_setup_kernel_pagetable(pgd_t *pgd, unsigned long max_pfn)
 {
 	pud_t *l3;
 	pmd_t *l2;
@@ -1781,8 +1780,6 @@ pgd_t * __init xen_setup_kernel_pagetable(pgd_t *pgd,
 
 	memblock_reserve(__pa(xen_start_info->pt_base),
 			 xen_start_info->nr_pt_frames * PAGE_SIZE);
-
-	return pgd;
 }
 #else	/* !CONFIG_X86_64 */
 static RESERVE_BRK_ARRAY(pmd_t, initial_kernel_pmd, PTRS_PER_PMD);
@@ -1825,8 +1822,7 @@ static void __init xen_write_cr3_init(unsigned long cr3)
 	pv_mmu_ops.write_cr3 = &xen_write_cr3;
 }
 
-pgd_t * __init xen_setup_kernel_pagetable(pgd_t *pgd,
-					 unsigned long max_pfn)
+void __init xen_setup_kernel_pagetable(pgd_t *pgd, unsigned long max_pfn)
 {
 	pmd_t *kernel_pmd;
 
@@ -1858,8 +1854,6 @@ pgd_t * __init xen_setup_kernel_pagetable(pgd_t *pgd,
 
 	memblock_reserve(__pa(xen_start_info->pt_base),
 			 xen_start_info->nr_pt_frames * PAGE_SIZE);
-
-	return initial_page_table;
 }
 #endif	/* CONFIG_X86_64 */
 
diff --git a/arch/x86/xen/xen-ops.h b/arch/x86/xen/xen-ops.h
index 202d4c1..2230f57 100644
--- a/arch/x86/xen/xen-ops.h
+++ b/arch/x86/xen/xen-ops.h
@@ -27,7 +27,7 @@ void xen_setup_mfn_list_list(void);
 void xen_setup_shared_info(void);
 void xen_build_mfn_list_list(void);
 void xen_setup_machphys_mapping(void);
-pgd_t *xen_setup_kernel_pagetable(pgd_t *pgd, unsigned long max_pfn);
+void xen_setup_kernel_pagetable(pgd_t *pgd, unsigned long max_pfn);
 void xen_reserve_top(void);
 extern unsigned long xen_max_p2m_pfn;
 
-- 
1.7.7.6

[PATCH 3/5] xen/mmu: Provide comments describing the _kva and _va aliasing issue

[Konrad Rzeszutek Wilk]

And also explain how the pagetables are being grafted on.
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
---
 arch/x86/xen/mmu.c |   15 +++++++++++++++
 1 files changed, 15 insertions(+), 0 deletions(-)

diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c
index 4ac21a4..b0b72e0 100644
--- a/arch/x86/xen/mmu.c
+++ b/arch/x86/xen/mmu.c
@@ -1734,19 +1734,34 @@ void __init xen_setup_kernel_pagetable(pgd_t *pgd, unsigned long max_pfn)
 	init_level4_pgt[0] = __pgd(0);
 
 	/* Pre-constructed entries are in pfn, so convert to mfn */
+	/* L4[272] -> level3_ident_pgt
+	 * L4[511] -> level3_kernel_pgt */
 	convert_pfn_mfn(init_level4_pgt);
+
+	/* L3_i[0] -> level2_ident_pgt */
 	convert_pfn_mfn(level3_ident_pgt);
+	/* L3_k[510] -> level2_kernel_pgt
+	 * L3_i[511] -> level2_fixmap_pgt */
 	convert_pfn_mfn(level3_kernel_pgt);
 
+	/* We get [511][511] and have Xen's version of level2_kernel_pgt */
 	l3 = m2v(pgd[pgd_index(__START_KERNEL_map)].pgd);
 	l2 = m2v(l3[pud_index(__START_KERNEL_map)].pud);
 
+	/* Graft it onto L4[272][0]. Note that we creating an aliasing problem:
+	 * Both L4[272][0] and L4[511][511] have entries that point to the same
+	 * L2 (PMD) tables. Meaning that if you modify it in __va space
+	 * it will be also modified in the __kva space! */
 	memcpy(level2_ident_pgt, l2, sizeof(pmd_t) * PTRS_PER_PMD);
+	/* Graft it onto L4[511][511] */
 	memcpy(level2_kernel_pgt, l2, sizeof(pmd_t) * PTRS_PER_PMD);
 
+	/* Get [511][510] and graft that in level2_fixmap_pgt */
 	l3 = m2v(pgd[pgd_index(__START_KERNEL_map + PMD_SIZE)].pgd);
 	l2 = m2v(l3[pud_index(__START_KERNEL_map + PMD_SIZE)].pud);
 	memcpy(level2_fixmap_pgt, l2, sizeof(pmd_t) * PTRS_PER_PMD);
+	/* Note that we don't do anything with level1_fixmap_pgt which
+	 * we don't need. */
 
 	/* Set up identity map */
 	xen_map_identity_early(level2_ident_pgt, max_pfn);
-- 
1.7.7.6

[PATCH 4/5] xen/x86: Use memblock_reserve for sensitive areas.

[Konrad Rzeszutek Wilk]

instead of a big memblock_reserve. This way we can be more
selective in freeing regions (and it also makes it easier
to understand where is what).

Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
---
 arch/x86/xen/enlighten.c |   36 ++++++++++++++++++++++++++++++++++++
 arch/x86/xen/p2m.c       |    5 +++++
 arch/x86/xen/setup.c     |    9 ---------
 3 files changed, 41 insertions(+), 9 deletions(-)

diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index ae8a2ab..c986b7f 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -998,7 +998,42 @@ static int xen_write_msr_safe(unsigned int msr, unsigned low, unsigned high)
 
 	return ret;
 }
+static void __init xen_reserve_mfn(unsigned long mfn)
+{
+	unsigned long pfn;
 
+	if (!mfn)
+		return;
+	pfn = mfn_to_pfn(mfn);
+	if (phys_to_machine_mapping_valid(pfn))
+		memblock_reserve(PFN_PHYS(pfn), PAGE_SIZE);
+}
+static void __init xen_reserve_internals(void)
+{
+	unsigned long size;
+
+	if (!xen_pv_domain())
+		return;
+
+	if (xen_feature(XENFEAT_auto_translated_physmap))
+		return;
+
+	/* ALIGN up to compensate for the p2m_page pointing to an array that
+	 * can be partially filled (look in xen_build_dynamic_phys_to_machine).
+	 */
+	size = PAGE_ALIGN(xen_start_info->nr_pages * sizeof(unsigned long));
+
+	memblock_reserve(__pa(xen_start_info->mfn_list), size);
+
+	memblock_reserve(__pa(xen_start_info), PAGE_SIZE);
+
+	xen_reserve_mfn(PFN_DOWN(xen_start_info->shared_info));
+	xen_reserve_mfn(xen_start_info->store_mfn);
+
+	if (!xen_initial_domain())
+		xen_reserve_mfn(xen_start_info->console.domU.mfn);
+	/* The pagetables are reserved in mmu.c */
+}
 void xen_setup_shared_info(void)
 {
 	if (!xen_feature(XENFEAT_auto_translated_physmap)) {
@@ -1357,6 +1392,7 @@ asmlinkage void __init xen_start_kernel(void)
 	xen_raw_console_write("mapping kernel into physical memory\n");
 	xen_setup_kernel_pagetable((pgd_t *)xen_start_info->pt_base, xen_start_info->nr_pages);
 
+	xen_reserve_internals();
 	/* Allocate and initialize top and mid mfn levels for p2m structure */
 	xen_build_mfn_list_list();
 
diff --git a/arch/x86/xen/p2m.c b/arch/x86/xen/p2m.c
index e4adbfb..4219f9a 100644
--- a/arch/x86/xen/p2m.c
+++ b/arch/x86/xen/p2m.c
@@ -388,6 +388,11 @@ void __init xen_build_dynamic_phys_to_machine(void)
 	}
 
 	m2p_override_init();
+
+	/* NOTE: We cannot call memblock_reserve here for the mfn_list as there
+	 * isn't enough pieces to make it work (for one - we are still using the
+	 * Xen provided pagetable). So we do it a bit later:
+	 * (xen_reserve_internals).*/
 }
 
 unsigned long get_phys_to_machine(unsigned long pfn)
diff --git a/arch/x86/xen/setup.c b/arch/x86/xen/setup.c
index a4790bf..9efca75 100644
--- a/arch/x86/xen/setup.c
+++ b/arch/x86/xen/setup.c
@@ -424,15 +424,6 @@ char * __init xen_memory_setup(void)
 	e820_add_region(ISA_START_ADDRESS, ISA_END_ADDRESS - ISA_START_ADDRESS,
 			E820_RESERVED);
 
-	/*
-	 * Reserve Xen bits:
-	 *  - mfn_list
-	 *  - xen_start_info
-	 * See comment above "struct start_info" in <xen/interface/xen.h>
-	 */
-	memblock_reserve(__pa(xen_start_info->mfn_list),
-			 xen_start_info->pt_base - xen_start_info->mfn_list);
-
 	sanitize_e820_map(e820.map, ARRAY_SIZE(e820.map), &e820.nr_map);
 
 	return "Xen";
-- 
1.7.7.6

[PATCH 5/5] xen/perf: Define .glob for the different hypercalls.

[Konrad Rzeszutek Wilk]

This allows us in perf to have this:

 99.67%  [kernel]             [k] xen_hypercall_sched_op
  0.11%  [kernel]             [k] xen_hypercall_xen_version

instead of the borring ever-encompassing:

 99.13%  [kernel]              [k] hypercall_page

[v2: Use a macro to define the name and skip]
[v3: Use balign per Jan's suggestion]
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
---
 arch/x86/xen/xen-head.S |   56 +++++++++++++++++++++++++++++++++++++++++++++-
 1 files changed, 54 insertions(+), 2 deletions(-)

diff --git a/arch/x86/xen/xen-head.S b/arch/x86/xen/xen-head.S
index aaa7291..7faed58 100644
--- a/arch/x86/xen/xen-head.S
+++ b/arch/x86/xen/xen-head.S
@@ -28,9 +28,61 @@ ENTRY(startup_xen)
 	__FINIT
 
 .pushsection .text
-	.align PAGE_SIZE
+	.balign PAGE_SIZE
 ENTRY(hypercall_page)
-	.skip PAGE_SIZE
+#define NEXT_HYPERCALL(x) \
+	ENTRY(xen_hypercall_##x) \
+	.skip 32
+
+NEXT_HYPERCALL(set_trap_table)
+NEXT_HYPERCALL(mmu_update)
+NEXT_HYPERCALL(set_gdt)
+NEXT_HYPERCALL(stack_switch)
+NEXT_HYPERCALL(set_callbacks)
+NEXT_HYPERCALL(fpu_taskswitch)
+NEXT_HYPERCALL(sched_op_compat)
+NEXT_HYPERCALL(platform_op)
+NEXT_HYPERCALL(set_debugreg)
+NEXT_HYPERCALL(get_debugreg)
+NEXT_HYPERCALL(update_descriptor)
+NEXT_HYPERCALL(ni)
+NEXT_HYPERCALL(memory_op)
+NEXT_HYPERCALL(multicall)
+NEXT_HYPERCALL(update_va_mapping)
+NEXT_HYPERCALL(set_timer_op)
+NEXT_HYPERCALL(event_channel_op_compat)
+NEXT_HYPERCALL(xen_version)
+NEXT_HYPERCALL(console_io)
+NEXT_HYPERCALL(physdev_op_compat)
+NEXT_HYPERCALL(grant_table_op)
+NEXT_HYPERCALL(vm_assist)
+NEXT_HYPERCALL(update_va_mapping_otherdomain)
+NEXT_HYPERCALL(iret)
+NEXT_HYPERCALL(vcpu_op)
+NEXT_HYPERCALL(set_segment_base)
+NEXT_HYPERCALL(mmuext_op)
+NEXT_HYPERCALL(xsm_op)
+NEXT_HYPERCALL(nmi_op)
+NEXT_HYPERCALL(sched_op)
+NEXT_HYPERCALL(callback_op)
+NEXT_HYPERCALL(xenoprof_op)
+NEXT_HYPERCALL(event_channel_op)
+NEXT_HYPERCALL(physdev_op)
+NEXT_HYPERCALL(hvm_op)
+NEXT_HYPERCALL(sysctl)
+NEXT_HYPERCALL(domctl)
+NEXT_HYPERCALL(kexec_op)
+NEXT_HYPERCALL(tmem_op) /* 38 */
+ENTRY(xen_hypercall_rsvr)
+	.skip 320
+NEXT_HYPERCALL(mca) /* 48 */
+NEXT_HYPERCALL(arch_1)
+NEXT_HYPERCALL(arch_2)
+NEXT_HYPERCALL(arch_3)
+NEXT_HYPERCALL(arch_4)
+NEXT_HYPERCALL(arch_5)
+NEXT_HYPERCALL(arch_6)
+	.balign PAGE_SIZE
 .popsection
 
 	ELFNOTE(Xen, XEN_ELFNOTE_GUEST_OS,       .asciz "linux")
-- 
1.7.7.6

Re: [Xen-devel] [PATCH 4/5] xen/x86: Use memblock_reserve for sensitive areas.

[Konrad Rzeszutek Wilk]

On Mon, Jul 23, 2012 at 02:28:22PM -0400, Konrad Rzeszutek Wilk wrote:
> instead of a big memblock_reserve. This way we can be more
> selective in freeing regions (and it also makes it easier
> to understand where is what).
> 
> Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
> ---
>  arch/x86/xen/enlighten.c |   36 ++++++++++++++++++++++++++++++++++++
>  arch/x86/xen/p2m.c       |    5 +++++
>  arch/x86/xen/setup.c     |    9 ---------
>  3 files changed, 41 insertions(+), 9 deletions(-)
> 
> diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
> index ae8a2ab..c986b7f 100644
> --- a/arch/x86/xen/enlighten.c
> +++ b/arch/x86/xen/enlighten.c
> @@ -998,7 +998,42 @@ static int xen_write_msr_safe(unsigned int msr, unsigned low, unsigned high)
>  
>  	return ret;
>  }
> +static void __init xen_reserve_mfn(unsigned long mfn)
> +{
> +	unsigned long pfn;
>  
> +	if (!mfn)
> +		return;
> +	pfn = mfn_to_pfn(mfn);
> +	if (phys_to_machine_mapping_valid(pfn))
> +		memblock_reserve(PFN_PHYS(pfn), PAGE_SIZE);
> +}
> +static void __init xen_reserve_internals(void)
> +{
> +	unsigned long size;
> +
> +	if (!xen_pv_domain())
> +		return;
> +
> +	if (xen_feature(XENFEAT_auto_translated_physmap))
> +		return;

This needs to be redone. On PV guests that do autotranslation the
other areas (xen_start_info for example) are not reserved  - which
is not good.
> +
> +	/* ALIGN up to compensate for the p2m_page pointing to an array that
> +	 * can be partially filled (look in xen_build_dynamic_phys_to_machine).
> +	 */
> +	size = PAGE_ALIGN(xen_start_info->nr_pages * sizeof(unsigned long));
> +
> +	memblock_reserve(__pa(xen_start_info->mfn_list), size);
> +
> +	memblock_reserve(__pa(xen_start_info), PAGE_SIZE);
> +
> +	xen_reserve_mfn(PFN_DOWN(xen_start_info->shared_info));
> +	xen_reserve_mfn(xen_start_info->store_mfn);
> +
> +	if (!xen_initial_domain())
> +		xen_reserve_mfn(xen_start_info->console.domU.mfn);
> +	/* The pagetables are reserved in mmu.c */
> +}
>  void xen_setup_shared_info(void)
>  {
>  	if (!xen_feature(XENFEAT_auto_translated_physmap)) {
> @@ -1357,6 +1392,7 @@ asmlinkage void __init xen_start_kernel(void)
>  	xen_raw_console_write("mapping kernel into physical memory\n");
>  	xen_setup_kernel_pagetable((pgd_t *)xen_start_info->pt_base, xen_start_info->nr_pages);
>  
> +	xen_reserve_internals();
>  	/* Allocate and initialize top and mid mfn levels for p2m structure */
>  	xen_build_mfn_list_list();
>  
> diff --git a/arch/x86/xen/p2m.c b/arch/x86/xen/p2m.c
> index e4adbfb..4219f9a 100644
> --- a/arch/x86/xen/p2m.c
> +++ b/arch/x86/xen/p2m.c
> @@ -388,6 +388,11 @@ void __init xen_build_dynamic_phys_to_machine(void)
>  	}
>  
>  	m2p_override_init();
> +
> +	/* NOTE: We cannot call memblock_reserve here for the mfn_list as there
> +	 * isn't enough pieces to make it work (for one - we are still using the
> +	 * Xen provided pagetable). So we do it a bit later:
> +	 * (xen_reserve_internals).*/
>  }
>  
>  unsigned long get_phys_to_machine(unsigned long pfn)
> diff --git a/arch/x86/xen/setup.c b/arch/x86/xen/setup.c
> index a4790bf..9efca75 100644
> --- a/arch/x86/xen/setup.c
> +++ b/arch/x86/xen/setup.c
> @@ -424,15 +424,6 @@ char * __init xen_memory_setup(void)
>  	e820_add_region(ISA_START_ADDRESS, ISA_END_ADDRESS - ISA_START_ADDRESS,
>  			E820_RESERVED);
>  
> -	/*
> -	 * Reserve Xen bits:
> -	 *  - mfn_list
> -	 *  - xen_start_info
> -	 * See comment above "struct start_info" in <xen/interface/xen.h>
> -	 */
> -	memblock_reserve(__pa(xen_start_info->mfn_list),
> -			 xen_start_info->pt_base - xen_start_info->mfn_list);
> -
>  	sanitize_e820_map(e820.map, ARRAY_SIZE(e820.map), &e820.nr_map);
>  
>  	return "Xen";
> -- 
> 1.7.7.6
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel