* [kvm-unit-tests PATCH v2 0/4] More lib/alloc cleanup and a minor improvement
@ 2020-07-06 16:43 Claudio Imbrenda
2020-07-06 16:43 ` [kvm-unit-tests PATCH v2 1/4] lib/vmalloc: fix pages count local variable to be size_t Claudio Imbrenda
` (4 more replies)
0 siblings, 5 replies; 9+ messages in thread
From: Claudio Imbrenda @ 2020-07-06 16:43 UTC (permalink / raw)
To: kvm, pbonzini; +Cc: frankja, thuth, david, drjones
Some more cleanup of lib/alloc in light of upcoming changes
The first real feature: allow aligned virtual allocations with
alignment greater than one page.
Also export a function for allocating aligned non-backed virtual pages.
v1->v2
* rename helper function to alloc_vpages_aligned, call it directly
* alloc_vpages_aligned now expects a page order as alignment
Claudio Imbrenda (4):
lib/vmalloc: fix pages count local variable to be size_t
lib/alloc_page: change some parameter types
lib/alloc_page: move get_order and is_power_of_2 to a bitops.h
lib/vmalloc: allow vm_memalign with alignment > PAGE_SIZE
lib/alloc_page.h | 7 +++----
lib/bitops.h | 10 ++++++++++
lib/libcflat.h | 5 -----
lib/vmalloc.h | 3 +++
lib/alloc.c | 1 +
lib/alloc_page.c | 13 ++++---------
lib/vmalloc.c | 37 ++++++++++++++++++++++++++++---------
7 files changed, 49 insertions(+), 27 deletions(-)
--
2.26.2
^ permalink raw reply [flat|nested] 9+ messages in thread
* [kvm-unit-tests PATCH v2 1/4] lib/vmalloc: fix pages count local variable to be size_t
2020-07-06 16:43 [kvm-unit-tests PATCH v2 0/4] More lib/alloc cleanup and a minor improvement Claudio Imbrenda
@ 2020-07-06 16:43 ` Claudio Imbrenda
2020-07-06 16:57 ` Jim Mattson
2020-07-06 16:43 ` [kvm-unit-tests PATCH v2 2/4] lib/alloc_page: change some parameter types Claudio Imbrenda
` (3 subsequent siblings)
4 siblings, 1 reply; 9+ messages in thread
From: Claudio Imbrenda @ 2020-07-06 16:43 UTC (permalink / raw)
To: kvm, pbonzini; +Cc: frankja, thuth, david, drjones
Since size is of type size_t, size >> PAGE_SHIFT might still be too big
for a normal unsigned int.
Signed-off-by: Claudio Imbrenda <imbrenda@linux.ibm.com>
Reviewed-by: Andrew Jones <drjones@redhat.com>
---
lib/vmalloc.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/lib/vmalloc.c b/lib/vmalloc.c
index 10f15af..9237a0f 100644
--- a/lib/vmalloc.c
+++ b/lib/vmalloc.c
@@ -40,7 +40,7 @@ void *alloc_vpage(void)
void *vmap(phys_addr_t phys, size_t size)
{
void *mem, *p;
- unsigned pages;
+ size_t pages;
size = PAGE_ALIGN(size);
pages = size / PAGE_SIZE;
@@ -58,7 +58,7 @@ void *vmap(phys_addr_t phys, size_t size)
static void *vm_memalign(size_t alignment, size_t size)
{
void *mem, *p;
- unsigned pages;
+ size_t pages;
assert(alignment <= PAGE_SIZE);
size = PAGE_ALIGN(size);
--
2.26.2
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [kvm-unit-tests PATCH v2 2/4] lib/alloc_page: change some parameter types
2020-07-06 16:43 [kvm-unit-tests PATCH v2 0/4] More lib/alloc cleanup and a minor improvement Claudio Imbrenda
2020-07-06 16:43 ` [kvm-unit-tests PATCH v2 1/4] lib/vmalloc: fix pages count local variable to be size_t Claudio Imbrenda
@ 2020-07-06 16:43 ` Claudio Imbrenda
2020-07-06 16:58 ` Jim Mattson
2020-07-13 15:16 ` Thomas Huth
2020-07-06 16:43 ` [kvm-unit-tests PATCH v2 3/4] lib/alloc_page: move get_order and is_power_of_2 to a bitops.h Claudio Imbrenda
` (2 subsequent siblings)
4 siblings, 2 replies; 9+ messages in thread
From: Claudio Imbrenda @ 2020-07-06 16:43 UTC (permalink / raw)
To: kvm, pbonzini; +Cc: frankja, thuth, david, drjones
For size parameters, size_t is probably semantically more appropriate
than unsigned long (although they map to the same value).
For order, unsigned long is just too big. Also, get_order returns an
unsigned int anyway.
Signed-off-by: Claudio Imbrenda <imbrenda@linux.ibm.com>
Reviewed-by: Andrew Jones <drjones@redhat.com>
---
lib/alloc_page.h | 6 +++---
lib/alloc_page.c | 8 ++++----
2 files changed, 7 insertions(+), 7 deletions(-)
diff --git a/lib/alloc_page.h b/lib/alloc_page.h
index 6181299..d9aceb7 100644
--- a/lib/alloc_page.h
+++ b/lib/alloc_page.h
@@ -11,10 +11,10 @@
bool page_alloc_initialized(void);
void page_alloc_ops_enable(void);
void *alloc_page(void);
-void *alloc_pages(unsigned long order);
+void *alloc_pages(unsigned int order);
void free_page(void *page);
-void free_pages(void *mem, unsigned long size);
-void free_pages_by_order(void *mem, unsigned long order);
+void free_pages(void *mem, size_t size);
+void free_pages_by_order(void *mem, unsigned int order);
unsigned int get_order(size_t size);
#endif
diff --git a/lib/alloc_page.c b/lib/alloc_page.c
index 8769c3f..f16eaad 100644
--- a/lib/alloc_page.c
+++ b/lib/alloc_page.c
@@ -21,7 +21,7 @@ bool page_alloc_initialized(void)
return freelist != 0;
}
-void free_pages(void *mem, unsigned long size)
+void free_pages(void *mem, size_t size)
{
void *old_freelist;
void *end;
@@ -53,7 +53,7 @@ void free_pages(void *mem, unsigned long size)
spin_unlock(&lock);
}
-void free_pages_by_order(void *mem, unsigned long order)
+void free_pages_by_order(void *mem, unsigned int order)
{
free_pages(mem, 1ul << (order + PAGE_SHIFT));
}
@@ -79,7 +79,7 @@ void *alloc_page()
* Allocates (1 << order) physically contiguous and naturally aligned pages.
* Returns NULL if there's no memory left.
*/
-void *alloc_pages(unsigned long order)
+void *alloc_pages(unsigned int order)
{
/* Generic list traversal. */
void *prev;
@@ -150,7 +150,7 @@ void free_page(void *page)
static void *page_memalign(size_t alignment, size_t size)
{
unsigned long n = ALIGN(size, PAGE_SIZE) >> PAGE_SHIFT;
- unsigned long order;
+ unsigned int order;
if (!size)
return NULL;
--
2.26.2
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [kvm-unit-tests PATCH v2 3/4] lib/alloc_page: move get_order and is_power_of_2 to a bitops.h
2020-07-06 16:43 [kvm-unit-tests PATCH v2 0/4] More lib/alloc cleanup and a minor improvement Claudio Imbrenda
2020-07-06 16:43 ` [kvm-unit-tests PATCH v2 1/4] lib/vmalloc: fix pages count local variable to be size_t Claudio Imbrenda
2020-07-06 16:43 ` [kvm-unit-tests PATCH v2 2/4] lib/alloc_page: change some parameter types Claudio Imbrenda
@ 2020-07-06 16:43 ` Claudio Imbrenda
2020-07-06 16:43 ` [kvm-unit-tests PATCH v2 4/4] lib/vmalloc: allow vm_memalign with alignment > PAGE_SIZE Claudio Imbrenda
2020-07-06 16:48 ` [kvm-unit-tests PATCH v2 0/4] More lib/alloc cleanup and a minor improvement Paolo Bonzini
4 siblings, 0 replies; 9+ messages in thread
From: Claudio Imbrenda @ 2020-07-06 16:43 UTC (permalink / raw)
To: kvm, pbonzini; +Cc: frankja, thuth, david, drjones
The functions get_order and is_power_of_2 are simple and should
probably be in a header, like similar simple functions in bitops.h
Since they concern bit manipulation, the logical place for them is in
bitops.h
Signed-off-by: Claudio Imbrenda <imbrenda@linux.ibm.com>
Reviewed-by: Andrew Jones <drjones@redhat.com>
---
lib/alloc_page.h | 1 -
lib/bitops.h | 10 ++++++++++
lib/libcflat.h | 5 -----
lib/alloc.c | 1 +
lib/alloc_page.c | 5 -----
5 files changed, 11 insertions(+), 11 deletions(-)
diff --git a/lib/alloc_page.h b/lib/alloc_page.h
index d9aceb7..88540d1 100644
--- a/lib/alloc_page.h
+++ b/lib/alloc_page.h
@@ -15,6 +15,5 @@ void *alloc_pages(unsigned int order);
void free_page(void *page);
void free_pages(void *mem, size_t size);
void free_pages_by_order(void *mem, unsigned int order);
-unsigned int get_order(size_t size);
#endif
diff --git a/lib/bitops.h b/lib/bitops.h
index b310a22..308aa86 100644
--- a/lib/bitops.h
+++ b/lib/bitops.h
@@ -74,4 +74,14 @@ static inline unsigned long fls(unsigned long word)
}
#endif
+static inline bool is_power_of_2(unsigned long n)
+{
+ return n && !(n & (n - 1));
+}
+
+static inline unsigned int get_order(size_t size)
+{
+ return size ? fls(size) + !is_power_of_2(size) : 0;
+}
+
#endif
diff --git a/lib/libcflat.h b/lib/libcflat.h
index 7092af2..ec0f58b 100644
--- a/lib/libcflat.h
+++ b/lib/libcflat.h
@@ -147,11 +147,6 @@ do { \
} \
} while (0)
-static inline bool is_power_of_2(unsigned long n)
-{
- return n && !(n & (n - 1));
-}
-
/*
* One byte per bit, a ' between each group of 4 bits, and a null terminator.
*/
diff --git a/lib/alloc.c b/lib/alloc.c
index 6c89f98..9d89d24 100644
--- a/lib/alloc.c
+++ b/lib/alloc.c
@@ -1,4 +1,5 @@
#include "alloc.h"
+#include "bitops.h"
#include "asm/page.h"
#include "bitops.h"
diff --git a/lib/alloc_page.c b/lib/alloc_page.c
index f16eaad..fa3c527 100644
--- a/lib/alloc_page.c
+++ b/lib/alloc_page.c
@@ -175,8 +175,3 @@ void page_alloc_ops_enable(void)
{
alloc_ops = &page_alloc_ops;
}
-
-unsigned int get_order(size_t size)
-{
- return is_power_of_2(size) ? fls(size) : fls(size) + 1;
-}
--
2.26.2
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [kvm-unit-tests PATCH v2 4/4] lib/vmalloc: allow vm_memalign with alignment > PAGE_SIZE
2020-07-06 16:43 [kvm-unit-tests PATCH v2 0/4] More lib/alloc cleanup and a minor improvement Claudio Imbrenda
` (2 preceding siblings ...)
2020-07-06 16:43 ` [kvm-unit-tests PATCH v2 3/4] lib/alloc_page: move get_order and is_power_of_2 to a bitops.h Claudio Imbrenda
@ 2020-07-06 16:43 ` Claudio Imbrenda
2020-07-06 16:48 ` [kvm-unit-tests PATCH v2 0/4] More lib/alloc cleanup and a minor improvement Paolo Bonzini
4 siblings, 0 replies; 9+ messages in thread
From: Claudio Imbrenda @ 2020-07-06 16:43 UTC (permalink / raw)
To: kvm, pbonzini; +Cc: frankja, thuth, david, drjones
Allow allocating aligned virtual memory with alignment larger than only
one page.
Add a check that the backing pages were actually allocated.
Export the alloc_vpages_aligned function to allow users to allocate
non-backed aligned virtual addresses.
Signed-off-by: Claudio Imbrenda <imbrenda@linux.ibm.com>
Reviewed-by: Andrew Jones <drjones@redhat.com>
---
lib/vmalloc.h | 3 +++
lib/vmalloc.c | 35 +++++++++++++++++++++++++++--------
2 files changed, 30 insertions(+), 8 deletions(-)
diff --git a/lib/vmalloc.h b/lib/vmalloc.h
index 2b563f4..8b158f5 100644
--- a/lib/vmalloc.h
+++ b/lib/vmalloc.h
@@ -5,6 +5,9 @@
/* Allocate consecutive virtual pages (without backing) */
extern void *alloc_vpages(ulong nr);
+/* Allocate consecutive and aligned virtual pages (without backing) */
+extern void *alloc_vpages_aligned(ulong nr, unsigned int alignment_order);
+
/* Allocate one virtual page (without backing) */
extern void *alloc_vpage(void);
/* Set the top of the virtual address space */
diff --git a/lib/vmalloc.c b/lib/vmalloc.c
index 9237a0f..e0c7b6b 100644
--- a/lib/vmalloc.c
+++ b/lib/vmalloc.c
@@ -12,19 +12,28 @@
#include "alloc.h"
#include "alloc_phys.h"
#include "alloc_page.h"
+#include <bitops.h>
#include "vmalloc.h"
static struct spinlock lock;
static void *vfree_top = 0;
static void *page_root;
-void *alloc_vpages(ulong nr)
+/*
+ * Allocate a certain number of pages from the virtual address space (without
+ * physical backing).
+ *
+ * nr is the number of pages to allocate
+ * alignment_pages is the alignment of the allocation *in pages*
+ */
+void *alloc_vpages_aligned(ulong nr, unsigned int align_order)
{
uintptr_t ptr;
spin_lock(&lock);
ptr = (uintptr_t)vfree_top;
ptr -= PAGE_SIZE * nr;
+ ptr &= GENMASK_ULL(63, PAGE_SHIFT + align_order);
vfree_top = (void *)ptr;
spin_unlock(&lock);
@@ -32,6 +41,11 @@ void *alloc_vpages(ulong nr)
return (void *)ptr;
}
+void *alloc_vpages(ulong nr)
+{
+ return alloc_vpages_aligned(nr, 0);
+}
+
void *alloc_vpage(void)
{
return alloc_vpages(1);
@@ -55,17 +69,22 @@ void *vmap(phys_addr_t phys, size_t size)
return mem;
}
+/*
+ * Allocate virtual memory, with the specified minimum alignment.
+ */
static void *vm_memalign(size_t alignment, size_t size)
{
+ phys_addr_t pa;
void *mem, *p;
- size_t pages;
- assert(alignment <= PAGE_SIZE);
- size = PAGE_ALIGN(size);
- pages = size / PAGE_SIZE;
- mem = p = alloc_vpages(pages);
- while (pages--) {
- phys_addr_t pa = virt_to_phys(alloc_page());
+ assert(is_power_of_2(alignment));
+
+ size = PAGE_ALIGN(size) / PAGE_SIZE;
+ alignment = get_order(PAGE_ALIGN(alignment) / PAGE_SIZE);
+ mem = p = alloc_vpages_aligned(size, alignment);
+ while (size--) {
+ pa = virt_to_phys(alloc_page());
+ assert(pa);
install_page(page_root, pa, p);
p += PAGE_SIZE;
}
--
2.26.2
^ permalink raw reply related [flat|nested] 9+ messages in thread
* Re: [kvm-unit-tests PATCH v2 0/4] More lib/alloc cleanup and a minor improvement
2020-07-06 16:43 [kvm-unit-tests PATCH v2 0/4] More lib/alloc cleanup and a minor improvement Claudio Imbrenda
` (3 preceding siblings ...)
2020-07-06 16:43 ` [kvm-unit-tests PATCH v2 4/4] lib/vmalloc: allow vm_memalign with alignment > PAGE_SIZE Claudio Imbrenda
@ 2020-07-06 16:48 ` Paolo Bonzini
4 siblings, 0 replies; 9+ messages in thread
From: Paolo Bonzini @ 2020-07-06 16:48 UTC (permalink / raw)
To: Claudio Imbrenda, kvm; +Cc: frankja, thuth, david, drjones
On 06/07/20 18:43, Claudio Imbrenda wrote:
> Some more cleanup of lib/alloc in light of upcoming changes
>
> The first real feature: allow aligned virtual allocations with
> alignment greater than one page.
>
> Also export a function for allocating aligned non-backed virtual pages.
>
> v1->v2
> * rename helper function to alloc_vpages_aligned, call it directly
> * alloc_vpages_aligned now expects a page order as alignment
>
> Claudio Imbrenda (4):
> lib/vmalloc: fix pages count local variable to be size_t
> lib/alloc_page: change some parameter types
> lib/alloc_page: move get_order and is_power_of_2 to a bitops.h
> lib/vmalloc: allow vm_memalign with alignment > PAGE_SIZE
>
> lib/alloc_page.h | 7 +++----
> lib/bitops.h | 10 ++++++++++
> lib/libcflat.h | 5 -----
> lib/vmalloc.h | 3 +++
> lib/alloc.c | 1 +
> lib/alloc_page.c | 13 ++++---------
> lib/vmalloc.c | 37 ++++++++++++++++++++++++++++---------
> 7 files changed, 49 insertions(+), 27 deletions(-)
>
Queued, thanks.
Paolo
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [kvm-unit-tests PATCH v2 1/4] lib/vmalloc: fix pages count local variable to be size_t
2020-07-06 16:43 ` [kvm-unit-tests PATCH v2 1/4] lib/vmalloc: fix pages count local variable to be size_t Claudio Imbrenda
@ 2020-07-06 16:57 ` Jim Mattson
0 siblings, 0 replies; 9+ messages in thread
From: Jim Mattson @ 2020-07-06 16:57 UTC (permalink / raw)
To: Claudio Imbrenda
Cc: kvm list, Paolo Bonzini, frankja, thuth, David Hildenbrand, drjones
On Mon, Jul 6, 2020 at 9:43 AM Claudio Imbrenda <imbrenda@linux.ibm.com> wrote:
>
> Since size is of type size_t, size >> PAGE_SHIFT might still be too big
> for a normal unsigned int.
>
> Signed-off-by: Claudio Imbrenda <imbrenda@linux.ibm.com>
> Reviewed-by: Andrew Jones <drjones@redhat.com>
Reviewed-by: Jim Mattson <jmattson@google.com>
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [kvm-unit-tests PATCH v2 2/4] lib/alloc_page: change some parameter types
2020-07-06 16:43 ` [kvm-unit-tests PATCH v2 2/4] lib/alloc_page: change some parameter types Claudio Imbrenda
@ 2020-07-06 16:58 ` Jim Mattson
2020-07-13 15:16 ` Thomas Huth
1 sibling, 0 replies; 9+ messages in thread
From: Jim Mattson @ 2020-07-06 16:58 UTC (permalink / raw)
To: Claudio Imbrenda
Cc: kvm list, Paolo Bonzini, frankja, thuth, David Hildenbrand, drjones
On Mon, Jul 6, 2020 at 9:43 AM Claudio Imbrenda <imbrenda@linux.ibm.com> wrote:
>
> For size parameters, size_t is probably semantically more appropriate
> than unsigned long (although they map to the same value).
>
> For order, unsigned long is just too big. Also, get_order returns an
> unsigned int anyway.
>
> Signed-off-by: Claudio Imbrenda <imbrenda@linux.ibm.com>
> Reviewed-by: Andrew Jones <drjones@redhat.com>
Reviewed-by: Jim Mattson <jmattson@google.com>
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [kvm-unit-tests PATCH v2 2/4] lib/alloc_page: change some parameter types
2020-07-06 16:43 ` [kvm-unit-tests PATCH v2 2/4] lib/alloc_page: change some parameter types Claudio Imbrenda
2020-07-06 16:58 ` Jim Mattson
@ 2020-07-13 15:16 ` Thomas Huth
1 sibling, 0 replies; 9+ messages in thread
From: Thomas Huth @ 2020-07-13 15:16 UTC (permalink / raw)
To: Claudio Imbrenda, kvm, pbonzini; +Cc: frankja, david, drjones
On 06/07/2020 18.43, Claudio Imbrenda wrote:
> For size parameters, size_t is probably semantically more appropriate
> than unsigned long (although they map to the same value).
>
> For order, unsigned long is just too big. Also, get_order returns an
> unsigned int anyway.
>
> Signed-off-by: Claudio Imbrenda <imbrenda@linux.ibm.com>
> Reviewed-by: Andrew Jones <drjones@redhat.com>
> ---
> lib/alloc_page.h | 6 +++---
> lib/alloc_page.c | 8 ++++----
> 2 files changed, 7 insertions(+), 7 deletions(-)
>
> diff --git a/lib/alloc_page.h b/lib/alloc_page.h
> index 6181299..d9aceb7 100644
> --- a/lib/alloc_page.h
> +++ b/lib/alloc_page.h
> @@ -11,10 +11,10 @@
> bool page_alloc_initialized(void);
> void page_alloc_ops_enable(void);
> void *alloc_page(void);
> -void *alloc_pages(unsigned long order);
> +void *alloc_pages(unsigned int order);
> void free_page(void *page);
> -void free_pages(void *mem, unsigned long size);
> -void free_pages_by_order(void *mem, unsigned long order);
> +void free_pages(void *mem, size_t size);
> +void free_pages_by_order(void *mem, unsigned int order);
> unsigned int get_order(size_t size);
>
> #endif
> diff --git a/lib/alloc_page.c b/lib/alloc_page.c
> index 8769c3f..f16eaad 100644
> --- a/lib/alloc_page.c
> +++ b/lib/alloc_page.c
> @@ -21,7 +21,7 @@ bool page_alloc_initialized(void)
> return freelist != 0;
> }
>
> -void free_pages(void *mem, unsigned long size)
> +void free_pages(void *mem, size_t size)
Hi Claudio,
this patch broke 32-bit x86 and arm builds:
https://travis-ci.com/github/huth/kvm-unit-tests/jobs/360418977#L693
https://travis-ci.com/github/huth/kvm-unit-tests/jobs/360418980#L545
I think you either need to adjust the format string, or cast the argument.
Thomas
^ permalink raw reply [flat|nested] 9+ messages in thread
end of thread, other threads:[~2020-07-13 15:17 UTC | newest]
Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-07-06 16:43 [kvm-unit-tests PATCH v2 0/4] More lib/alloc cleanup and a minor improvement Claudio Imbrenda
2020-07-06 16:43 ` [kvm-unit-tests PATCH v2 1/4] lib/vmalloc: fix pages count local variable to be size_t Claudio Imbrenda
2020-07-06 16:57 ` Jim Mattson
2020-07-06 16:43 ` [kvm-unit-tests PATCH v2 2/4] lib/alloc_page: change some parameter types Claudio Imbrenda
2020-07-06 16:58 ` Jim Mattson
2020-07-13 15:16 ` Thomas Huth
2020-07-06 16:43 ` [kvm-unit-tests PATCH v2 3/4] lib/alloc_page: move get_order and is_power_of_2 to a bitops.h Claudio Imbrenda
2020-07-06 16:43 ` [kvm-unit-tests PATCH v2 4/4] lib/vmalloc: allow vm_memalign with alignment > PAGE_SIZE Claudio Imbrenda
2020-07-06 16:48 ` [kvm-unit-tests PATCH v2 0/4] More lib/alloc cleanup and a minor improvement Paolo Bonzini
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).