libceph: odds and ends
cover libceph: odds and ends
(These patches, along with the patches I'll be posting in the next few minutes, are available in the "review/wip-stripe-v2" branch of the ceph-client git repository.)
This series just fleshes out some code in libceph. The first completes the set of definitions for limit values (like U8_MAX) to include the signed minimum and maximum values. The second just adds some error checking and type casting to timespec encoding and decoding routines.
-Alex
[PATCH 1/2] libceph: add signed type limits [PATCH 2/2] libceph: validate timespec conversions
PATCH 1/2 libceph: add signed type limits
A ceph timespec contains 32-bit unsigned values for its seconds and nanoseconds components. For a standard timespec, both fields are signed, and the seconds field is almost surely 64 bits.
Add some explicit casts so the fact that this conversion is taking place is obvious. Also trip a bug if we ever try to put out of range (negative or too big) values into a ceph timespec.
--- include/linux/ceph/decode.h | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-)
include/linux/ceph/decode.h +9 -4
@@ -154,14 +154,19 @@ bad:static inline void ceph_decode_timespec (struct timespec *ts,const struct ceph_timespec *tv) { - ts->tv_sec =le32_to_cpu (tv->tv_sec); - ts->tv_nsec =le32_to_cpu (tv->tv_nsec); + ts->tv_sec = (__kernel_time_t )le32_to_cpu (tv->tv_sec); + ts->tv_nsec = (long )le32_to_cpu (tv->tv_nsec); }static inline void ceph_encode_timespec (struct ceph_timespec *tv,const struct timespec *ts) { - tv->tv_sec =cpu_to_le32 (ts->tv_sec); - tv->tv_nsec =cpu_to_le32 (ts->tv_nsec); +BUG_ON (ts->tv_sec <0 ); +BUG_ON (ts->tv_sec > (__kernel_time_t )U32_MAX ); +BUG_ON (ts->tv_nsec <0 ); +BUG_ON (ts->tv_nsec > (long )U32_MAX ); + + tv->tv_sec =cpu_to_le32 ((u32 )ts->tv_sec); + tv->tv_nsec =cpu_to_le32 ((u32 )ts->tv_nsec); }/*
Signature · 1 line
1.7.9.5
Re: [PATCH 1/2] libceph: add signed type limits
(Sorry, the subject line didn't match the content. Please ignore the previous one--this is the right patch.)
Flesh out the limits defined in <linux/ceph/decode.h> to include the maximum and minimum values for signed type S8, S16, S32, and S64.
--- include/linux/ceph/decode.h | 18 ++++++++++++++---- 1 file changed, 14 insertions(+), 4 deletions(-)
include/linux/ceph/decode.h +14 -4
@@ -10,10 +10,20 @@/* This seemed to be the easiest place to define these */ -#define U8_MAX ((u8) (~0U)) -#define U16_MAX ((u16) (~0U)) -#define U32_MAX ((u32) (~0U)) -#define U64_MAX ((u64) (~0ULL)) +#define U8_MAX ((u8)(~0U)) +#define U16_MAX ((u16)(~0U)) +#define U32_MAX ((u32)(~0U)) +#define U64_MAX ((u64)(~0ULL)) + +#define S8_MAX ((s8)(U8_MAX >> 1)) +#define S16_MAX ((s16)(U16_MAX >> 1)) +#define S32_MAX ((s32)(U32_MAX >> 1)) +#define S64_MAX ((s64)(U64_MAX >> 1LL)) + +#define S8_MIN ((s8)(-S8_MAX - 1)) +#define S16_MIN ((s16)(-S16_MAX - 1)) +#define S32_MIN ((s32)(-S32_MAX - 1)) +#define S64_MIN ((s64)(-S64_MAX - 1LL))/* * in all cases,
Signature · 1 line
1.7.9.5
PATCH 2/2 libceph: validate timespec conversions
In reply to Alex ElderA ceph timespec contains 32-bit unsigned values for its seconds and nanoseconds components. For a standard timespec, both fields are signed, and the seconds field is almost surely 64 bits.
Add some explicit casts so the fact that this conversion is taking place is obvious. Also trip a bug if we ever try to put out of range (negative or too big) values into a ceph timespec.
--- include/linux/ceph/decode.h | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-)
include/linux/ceph/decode.h +9 -4
@@ -154,14 +154,19 @@ bad:static inline void ceph_decode_timespec (struct timespec *ts,const struct ceph_timespec *tv) { - ts->tv_sec =le32_to_cpu (tv->tv_sec); - ts->tv_nsec =le32_to_cpu (tv->tv_nsec); + ts->tv_sec = (__kernel_time_t )le32_to_cpu (tv->tv_sec); + ts->tv_nsec = (long )le32_to_cpu (tv->tv_nsec); }static inline void ceph_encode_timespec (struct ceph_timespec *tv,const struct timespec *ts) { - tv->tv_sec =cpu_to_le32 (ts->tv_sec); - tv->tv_nsec =cpu_to_le32 (ts->tv_nsec); +BUG_ON (ts->tv_sec <0 ); +BUG_ON (ts->tv_sec > (__kernel_time_t )U32_MAX ); +BUG_ON (ts->tv_nsec <0 ); +BUG_ON (ts->tv_nsec > (long )U32_MAX ); + + tv->tv_sec =cpu_to_le32 ((u32 )ts->tv_sec); + tv->tv_nsec =cpu_to_le32 ((u32 )ts->tv_nsec); }/*
Signature · 1 line
1.7.9.5
Re: [PATCH 2/2] libceph: validate timespec conversions
Quoted 3 lines · ----- "Alex Elder
A ceph timespec contains 32-bit unsigned values for its seconds and nanoseconds components. For a standard timespec, both fields are signed, and the seconds field is almost surely 64 bits.
Is the Ceph timespec going to change at some point?
Quoted 5 lines
Add some explicit casts so the fact that this conversion is taking place is obvious. Also trip a bug if we ever try to put out of range (negative or too big) values into a ceph timespec.
Signature · 10 lines
Matt Benjamin The Linux Box 206 South Fifth Ave. Suite 150 Ann Arbor, MI 48104 http://linuxbox.com tel. 734-761-4689 fax. 734-769-8938 cel. 734-216-5309
Re: [PATCH 2/2] libceph: validate timespec conversions
Quoted 8 lines · Matt W. Benjamin
Quoted 3 lines · ----- "Alex Elder
A ceph timespec contains 32-bit unsigned values for its seconds and nanoseconds components. For a standard timespec, both fields are signed, and the seconds field is almost surely 64 bits.
Is the Ceph timespec going to change at some point?
I don't think so. 32-bits is enough for the billion nanoseconds in a second. And I'm not sure if the signedness is used/useful... the ceph utime_t code always normalizes the ns result to be in [0, 1 billion).
sage
Quoted 24 lines
Quoted 6 lines
Add some explicit casts so the fact that this conversion is taking place is obvious. Also trip a bug if we ever try to put out of range (negative or too big) values into a ceph timespec.
Signature · 14 lines
Matt Benjamin The Linux Box 206 South Fifth Ave. Suite 150 Ann Arbor, MI 48104 http://linuxbox.com tel. 734-761-4689 fax. 734-769-8938 cel. 734-216-5309 -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 2/2] libceph: validate timespec conversions
I was thinking about the seconds component.
Quoted 18 lines · ----- "Sage Weil
Quoted 4 lines · Matt W. Benjamin
Quoted 1 line · ----- "Alex Elder
A ceph timespec contains 32-bit unsigned values for its seconds
and
Quoted 4 lines
Quoted 2 lines
nanoseconds components. For a standard timespec, both fields are signed, and the seconds field is almost surely 64 bits.
Is the Ceph timespec going to change at some point?
I don't think so. 32-bits is enough for the billion nanoseconds in a
second. And I'm not sure if the signedness is used/useful... the ceph
utime_tcode always normalizes the ns result to be in [0, 1 billion).sage
Signature · 10 lines
Matt Benjamin The Linux Box 206 South Fifth Ave. Suite 150 Ann Arbor, MI 48104 http://linuxbox.com tel. 734-761-4689 fax. 734-769-8938 cel. 734-216-5309
Re: [PATCH 2/2] libceph: validate timespec conversions
Quoted 1 line · Matt W. Benjamin
I was thinking about the seconds component.
Oh, right.. that's the unix epoch(alypse) in 2038 or something?
That we should probably fix. :)
s
Quoted 40 lines
Quoted 18 lines · ----- "Sage Weil
Quoted 4 lines · Matt W. Benjamin
Quoted 1 line · ----- "Alex Elder
A ceph timespec contains 32-bit unsigned values for its seconds
and
Quoted 4 lines
Quoted 2 lines
nanoseconds components. For a standard timespec, both fields are signed, and the seconds field is almost surely 64 bits.
Is the Ceph timespec going to change at some point?
I don't think so. 32-bits is enough for the billion nanoseconds in a
second. And I'm not sure if the signedness is used/useful... the ceph
utime_tcode always normalizes the ns result to be in [0, 1 billion).sage
Signature · 14 lines
Matt Benjamin The Linux Box 206 South Fifth Ave. Suite 150 Ann Arbor, MI 48104 http://linuxbox.com tel. 734-761-4689 fax. 734-769-8938 cel. 734-216-5309 -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 2/2] libceph: validate timespec conversions
In reply to Matt W. BenjaminQuoted 1 line · Matt W. Benjamin
I was thinking about the seconds component.
I wondered the same thing. It will most likely
have to some time in the next 25 years or so.
-Alex
Quoted 22 lines
Quoted 18 lines · ----- "Sage Weil
Quoted 4 lines · Matt W. Benjamin
Quoted 1 line · ----- "Alex Elder
A ceph timespec contains 32-bit unsigned values for its seconds
and
Quoted 4 lines
Quoted 2 lines
nanoseconds components. For a standard timespec, both fields are signed, and the seconds field is almost surely 64 bits.
Is the Ceph timespec going to change at some point?
I don't think so. 32-bits is enough for the billion nanoseconds in a
second. And I'm not sure if the signedness is used/useful... the ceph
utime_tcode always normalizes the ns result to be in [0, 1 billion).sage
Josh Durgin
Reviewed-byRe: [PATCH 1/2] libceph: add signed type limits
In reply to Alex ElderQuoted 43 lines · Alex Elder
(Sorry, the subject line didn't match the content. Please ignore the previous one--this is the right patch.)
Flesh out the limits defined in <
linux/ceph/decode.h> to include the maximum and minimum values for signed type S8, S16, S32, and S64.Signed-off-by Alex Elder <[email protected]>--- include/linux/ceph/decode.h | 18 ++++++++++++++---- 1 file changed, 14 insertions(+), 4 deletions(-)include/linux/ceph/decode.h +14
-4@@ -10,10 +10,20 @@/* This seemed to be the easiest place to define these */ -#define U8_MAX ((u8) (~0U)) -#define U16_MAX ((u16) (~0U)) -#define U32_MAX ((u32) (~0U)) -#define U64_MAX ((u64) (~0ULL)) +#define U8_MAX ((u8)(~0U)) +#define U16_MAX ((u16)(~0U)) +#define U32_MAX ((u32)(~0U)) +#define U64_MAX ((u64)(~0ULL)) + +#define S8_MAX ((s8)(U8_MAX >> 1)) +#define S16_MAX ((s16)(U16_MAX >> 1)) +#define S32_MAX ((s32)(U32_MAX >> 1)) +#define S64_MAX ((s64)(U64_MAX >> 1LL)) + +#define S8_MIN ((s8)(-S8_MAX - 1)) +#define S16_MIN ((s16)(-S16_MAX - 1)) +#define S32_MIN ((s32)(-S32_MAX - 1)) +#define S64_MIN ((s64)(-S64_MAX - 1LL))/* * in all cases,
Re: [PATCH 2/2] libceph: validate timespec conversions
In reply to Alex ElderQuoted 42 lines · Alex Elder
A ceph timespec contains 32-bit unsigned values for its seconds and nanoseconds components. For a standard timespec, both fields are signed, and the seconds field is almost surely 64 bits.
Add some explicit casts so the fact that this conversion is taking place is obvious. Also trip a bug if we ever try to put out of range (negative or too big) values into a ceph timespec.
Signed-off-by Alex Elder <[email protected]>--- include/linux/ceph/decode.h | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-)include/linux/ceph/decode.h +9
-4@@ -154,14 +154,19 @@ bad:static inline void ceph_decode_timespec (struct timespec *ts,const struct ceph_timespec *tv) { - ts->tv_sec =le32_to_cpu (tv->tv_sec); - ts->tv_nsec =le32_to_cpu (tv->tv_nsec); + ts->tv_sec = (__kernel_time_t )le32_to_cpu (tv->tv_sec); + ts->tv_nsec = (long )le32_to_cpu (tv->tv_nsec); }static inline void ceph_encode_timespec (struct ceph_timespec *tv,const struct timespec *ts) { - tv->tv_sec =cpu_to_le32 (ts->tv_sec); - tv->tv_nsec =cpu_to_le32 (ts->tv_nsec); +BUG_ON (ts->tv_sec <0 ); +BUG_ON (ts->tv_sec > (__kernel_time_t )U32_MAX ); +BUG_ON (ts->tv_nsec <0 ); +BUG_ON (ts->tv_nsec > (long )U32_MAX ); + + tv->tv_sec =cpu_to_le32 ((u32 )ts->tv_sec); + tv->tv_nsec =cpu_to_le32 ((u32 )ts->tv_nsec); }/*