Re: [PATCH v2 0/4] VPHN parsing fixes
On Tue, 2015-02-03 at 08:46 +0100, Greg Kurz wrote: > On Tue, 03 Feb 2015 13:47:35 +1100 > Michael Ellerman wrote: > > > On Thu, 2015-01-29 at 19:03 +0100, Greg Kurz wrote: > > > On Wed, 17 Dec 2014 10:40:46 +0100 > > > Greg Kurz wrote: > > > > Hi, > > > > > > > > This series addresses remarks from Ben and Michael (see individual > > > > patches). > > > > The most notable changes are: > > > > - the parsing code being pull out into a separate file in patch 3/4. > > > > This > > > > allows to write userland tests like the one below. > > > > - a full rewrite of the parsing logic in patch 4/4 > > > > > > > > > > Ping ? > > > > Sorry, lots of patches needing review. > > Heh, no problem. :) Since this isn't bugfix, I can even repost later, when > the review pressure is lower. Yeah if you can repost in ~2 weeks that would be perfect. > > This looks pretty good at a glance, but did you actually write a userspace > > test > > for it? If so please send it. I'm happy to rework it into something that > > can go > > in selftests. > > > > Yes, I wrote the premise of a test program. You can find it in the cover mail > of > this series: > > https://lists.ozlabs.org/pipermail/linuxppc-dev/2014-December/123601.html > > I'll have a look at selftests. Thanks. That test looks like a good start, there might be some more cases to test though? If you just copy the tools/testing/selftests/powerpc/tm directory and rename to vhpn, you can probably work out the rest. cheers ___ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
Re: [PATCH v2 0/4] VPHN parsing fixes
On Tue, 03 Feb 2015 13:47:35 +1100 Michael Ellerman wrote: > On Thu, 2015-01-29 at 19:03 +0100, Greg Kurz wrote: > > On Wed, 17 Dec 2014 10:40:46 +0100 > > Greg Kurz wrote: > > > Hi, > > > > > > This series addresses remarks from Ben and Michael (see individual > > > patches). > > > The most notable changes are: > > > - the parsing code being pull out into a separate file in patch 3/4. This > > > allows to write userland tests like the one below. > > > - a full rewrite of the parsing logic in patch 4/4 > > > > > > > Ping ? > > Sorry, lots of patches needing review. > Heh, no problem. :) Since this isn't bugfix, I can even repost later, when the review pressure is lower. > This looks pretty good at a glance, but did you actually write a userspace > test > for it? If so please send it. I'm happy to rework it into something that can > go > in selftests. > Yes, I wrote the premise of a test program. You can find it in the cover mail of this series: https://lists.ozlabs.org/pipermail/linuxppc-dev/2014-December/123601.html I'll have a look at selftests. > cheers > > Thanks. -- Greg ___ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
Re: [PATCH v2 0/4] VPHN parsing fixes
On Thu, 2015-01-29 at 19:03 +0100, Greg Kurz wrote: > On Wed, 17 Dec 2014 10:40:46 +0100 > Greg Kurz wrote: > > Hi, > > > > This series addresses remarks from Ben and Michael (see individual patches). > > The most notable changes are: > > - the parsing code being pull out into a separate file in patch 3/4. This > > allows to write userland tests like the one below. > > - a full rewrite of the parsing logic in patch 4/4 > > > > Ping ? Sorry, lots of patches needing review. This looks pretty good at a glance, but did you actually write a userspace test for it? If so please send it. I'm happy to rework it into something that can go in selftests. cheers ___ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
Re: [PATCH v2 0/4] VPHN parsing fixes
On Wed, 17 Dec 2014 10:40:46 +0100
Greg Kurz wrote:
> Hi,
>
> This series addresses remarks from Ben and Michael (see individual patches).
> The most notable changes are:
> - the parsing code being pull out into a separate file in patch 3/4. This
> allows to write userland tests like the one below.
> - a full rewrite of the parsing logic in patch 4/4
>
Ping ?
> --
> #include
> #include
>
> typedef unsigned long u64;
> typedef unsigned int u32;
> typedef unsigned short u16;
> typedef enum { false = 0, true } bool;
>
> #if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
> #define cpu_to_be32(x)bswap_32(x)
> #define be32_to_cpu(x)bswap_32(x)
> #define be16_to_cpup(x) bswap_16(*x)
> #define cpu_to_be64(x)bswap_64(x)
> #else
> #define cpu_to_be32(x)(x)
> #define be32_to_cpu(x)(x)
> #define be16_to_cpup(x) (*x)
> #define cpu_to_be64(x)(x)
> #endif
>
> #define pr_debug(...) printf(__VA_ARGS__)
>
> #include "vphn.c"
>
> void print_packed(const long *packed)
> {
> char *p = (char*) packed;
> int i;
>
> printf("\nRegisters:\n");
> for (i = 0; i < VPHN_REGISTER_COUNT; i++)
> printf("0x%016lx\n", packed[i]);
>
> printf("\nMemory layout:\n");
> for (i = 0; i < 6; i++) {
> printf("0x %02hhx %02hhx %02hhx %02hhx"
> " %02hhx %02hhx %02hhx %02hhx\n",
> *(p + 0), *(p + 1), *(p + 2), *(p + 3),
> *(p + 4), *(p + 5), *(p + 6), *(p + 7));
> p += 8;
> }
>
> putchar('\n');
> }
>
> void print_unpacked(const __be32 *unpacked)
> {
> int i;
>
> printf("\nVPHN associativity:\n");
> for (i = 0; i <= be32_to_cpu(unpacked[0]); i++)
> printf("0x%08x\n", be32_to_cpu(unpacked[i]));
>
> putchar('\n');
> }
>
> int main(int argc, char **argv)
> {
> int i;
> struct {
> const char *descr;
> long packed[VPHN_REGISTER_COUNT];
> } data[] = {
> {
> "16-bit and 32-bit",
> 0x8001800280038004,
> 0x8005800680078008,
> 0x0009000a,
> 0x000b000c,
> 0x,
> 0x
> },
> {
> "filled with 16-bit",
> 0x8001800280038004,
> 0x8005800680078008,
> 0x8009800a800b800c,
> 0x800d800e800f8010,
> 0x8011801280138014,
> 0x8015801680178018,
> },
> {
> "filled with 32-bit",
> 0x00010002,
> 0x00030004,
> 0x00050006,
> 0x00070008,
> 0x0009000a,
> 0x000b000c,
> },
> {
> "32-bit has all ones in 16 lower bits",
> 0x000180028003,
> 0x,
> 0x,
> 0x,
> 0x,
> 0x,
> },
> {
> "32-bit accross two 64-bit registers",
> 0x80010002,
> 0x000300048005,
> 0x,
> 0x,
> 0x,
> 0x,
> },
> {
> "Truncated last 32-bit",
> 0x00010002,
> 0x00030004,
> 0x00050006,
> 0x00070008,
> 0x0009000a,
> 0x000b800c0bad,
> },
> };
>
> for (i = 0; i < sizeof(data) / sizeof(data[0]); i++) {
> __be32 unpacked[VPHN_ASSOC_BUFSIZE] = { 0 };
>
> printf("\n==\n");
> printf("\nSet #%d: %s\n", i, data[i].descr);
>
> printf("\n==\n");
> print_packed(data[i].packed);
> vphn_unpack_associativity(data[i].packed, unpacked);
> print_unpacked(unpacked);
> }
>
> return 0;
> }
>
> ---
>
> Greg Kurz (4):
> powerpc/vphn: clarify the H_HOME_NODE_ASSOCIATIVITY API
> powerpc/vphn: move endianness fixing to vphn_unpack_associativity()
> powerpc/vphn: move VPHN parsing logic
[PATCH v2 0/4] VPHN parsing fixes
Hi,
This series addresses remarks from Ben and Michael (see individual patches).
The most notable changes are:
- the parsing code being pull out into a separate file in patch 3/4. This
allows to write userland tests like the one below.
- a full rewrite of the parsing logic in patch 4/4
--
#include
#include
typedef unsigned long u64;
typedef unsigned int u32;
typedef unsigned short u16;
typedef enum { false = 0, true } bool;
#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
#define cpu_to_be32(x) bswap_32(x)
#define be32_to_cpu(x) bswap_32(x)
#define be16_to_cpup(x) bswap_16(*x)
#define cpu_to_be64(x) bswap_64(x)
#else
#define cpu_to_be32(x) (x)
#define be32_to_cpu(x) (x)
#define be16_to_cpup(x) (*x)
#define cpu_to_be64(x) (x)
#endif
#define pr_debug(...) printf(__VA_ARGS__)
#include "vphn.c"
void print_packed(const long *packed)
{
char *p = (char*) packed;
int i;
printf("\nRegisters:\n");
for (i = 0; i < VPHN_REGISTER_COUNT; i++)
printf("0x%016lx\n", packed[i]);
printf("\nMemory layout:\n");
for (i = 0; i < 6; i++) {
printf("0x %02hhx %02hhx %02hhx %02hhx"
" %02hhx %02hhx %02hhx %02hhx\n",
*(p + 0), *(p + 1), *(p + 2), *(p + 3),
*(p + 4), *(p + 5), *(p + 6), *(p + 7));
p += 8;
}
putchar('\n');
}
void print_unpacked(const __be32 *unpacked)
{
int i;
printf("\nVPHN associativity:\n");
for (i = 0; i <= be32_to_cpu(unpacked[0]); i++)
printf("0x%08x\n", be32_to_cpu(unpacked[i]));
putchar('\n');
}
int main(int argc, char **argv)
{
int i;
struct {
const char *descr;
long packed[VPHN_REGISTER_COUNT];
} data[] = {
{
"16-bit and 32-bit",
0x8001800280038004,
0x8005800680078008,
0x0009000a,
0x000b000c,
0x,
0x
},
{
"filled with 16-bit",
0x8001800280038004,
0x8005800680078008,
0x8009800a800b800c,
0x800d800e800f8010,
0x8011801280138014,
0x8015801680178018,
},
{
"filled with 32-bit",
0x00010002,
0x00030004,
0x00050006,
0x00070008,
0x0009000a,
0x000b000c,
},
{
"32-bit has all ones in 16 lower bits",
0x000180028003,
0x,
0x,
0x,
0x,
0x,
},
{
"32-bit accross two 64-bit registers",
0x80010002,
0x000300048005,
0x,
0x,
0x,
0x,
},
{
"Truncated last 32-bit",
0x00010002,
0x00030004,
0x00050006,
0x00070008,
0x0009000a,
0x000b800c0bad,
},
};
for (i = 0; i < sizeof(data) / sizeof(data[0]); i++) {
__be32 unpacked[VPHN_ASSOC_BUFSIZE] = { 0 };
printf("\n==\n");
printf("\nSet #%d: %s\n", i, data[i].descr);
printf("\n==\n");
print_packed(data[i].packed);
vphn_unpack_associativity(data[i].packed, unpacked);
print_unpacked(unpacked);
}
return 0;
}
---
Greg Kurz (4):
powerpc/vphn: clarify the H_HOME_NODE_ASSOCIATIVITY API
powerpc/vphn: move endianness fixing to vphn_unpack_associativity()
powerpc/vphn: move VPHN parsing logic to a separate file
powerpc/vphn: parsing code rewrite
arch/powerpc/mm/Makefile |1 +
arch/powerpc/mm/numa.c | 55 ++--
arch/powerpc/mm/vphn.c | 70 ++
