scatterlist: add new capabilities
Subsystems (5)
v5 cover scatterlist: add new capabilities
Scatter-gather lists (sgl_s) are frequently used as data carriers in the block layer. For example the SCSI and NVMe subsystems interchange data with the block layer using sgl_s. The sgl API is declared in <linux/scatterlist.h>
The author has extended these transient sgl use cases to a store (i.e. a ramdisk) in the scsi_debug driver. Other new potential uses of sgl_s could be for the target subsystem. When this extra step is taken, the need to copy between sgl_s becomes apparent. The patchset adds sgl_copy_sgl(), sgl_compare_sgl() and sgl_memset().
The existing sgl_alloc_order() function can be seen as a replacement for vmalloc() for large, long-term allocations. For what seems like no good reason, sgl_alloc_order() currently restricts its total allocation to less than or equal to 4 GiB. vmalloc() has no such restriction.
Changes since v3 [posted 20201105]:
- rebase on lk 5.11.0-rc2
Changes since v3 [posted 20201019]:
- re-instate check on integer overflow of nent calculation in sgl_alloc_order(). Do it in such a way as to not limit the overall sgl size to 4 GiB
- introduce sgl_compare_sgl_idx() helper function that, if requested and if a miscompare is detected, will yield the byte index of the first miscompare.
- add Reviewed-by tags from Bodo Stroesser
- rebase on lk 5.10.0-rc2 [was on lk 5.9.0]
Changes since v2 [posted 20201018]:
- remove unneeded lines from sgl_memset() definition.
- change sg_zero_buffer() to call sgl_memset() as the former is a subset.
Changes since v1 [posted 20201016]:
- Bodo Stroesser pointed out a problem with the nesting of kmap_atomic() [called via sg_miter_next()] and kunmap_atomic() calls [called via sg_miter_stop()] and proposed a solution that simplifies the previous code.
- the new implementation of the three functions has shorter periods when pre-emption is disabled (but has more them). This should make operations on large sgl_s more pre-emption "friendly" with a relatively small performance hit.
- sgl_memset return type changed from void to size_t and is the number of bytes actually (over)written. That number is needed anyway internally so may as well return it as it may be useful to the caller.
This patchset is against lk 5.10.0-rc2
Douglas Gilbert (4):
sgl_alloc_order: remove 4 GiB limit, sgl_free() warning
scatterlist: add sgl_copy_sgl() function
scatterlist: add sgl_compare_sgl() function
scatterlist: add sgl_memset()
include/linux/scatterlist.h | 16 +++ lib/scatterlist.c | 244 +++++++++++++++++++++++++++++++++--- 2 files changed, 243 insertions(+), 17 deletions(-)
Signature · 1 line
2.25.1
PATCH v5 1/4 sgl_alloc_order: remove 4 GiB limit, sgl_free() warning
This patch fixes a check done by sgl_alloc_order() before it starts any allocations. The comment in the original said: "Check for integer overflow" but the check itself contained an integer overflow! The right hand side (rhs) of the expression in the condition is resolved as u32 so it could not exceed UINT32_MAX (4 GiB) which means 'length' could not exceed that value. If that was the intention then the comment above it could be dropped and the condition rewritten more clearly as:
if (length > UINT32_MAX) <<failure path >>;
Get around the integer overflow problem in the rhs of the original check by taking ilog2() of both sides.
This function may be used to replace vmalloc(unsigned long) for a large allocation (e.g. a ramdisk). vmalloc has no limit at 4 GiB so it seems unreasonable that:
sgl_alloc_order(unsigned long long length, ....)
does. sgl_s made with sgl_alloc_order() have equally sized segments placed in a scatter gather array. That allows O(1) navigation around a big sgl using some simple integer arithmetic.
Revise some of this function's description to more accurately reflect what this function is doing.
An earlier patch fixed a memory leak in sg_alloc_order() due to the misuse of sgl_free(). Take the opportunity to put a one line comment above sgl_free()'s declaration warning that it is not suitable when order > 0 .
--- include/linux/scatterlist.h | 1 + lib/scatterlist.c | 14 ++++++++------ 2 files changed, 9 insertions(+), 6 deletions(-)
include/linux/scatterlist.h +1 -0
@@ -302,6 +302,7 @@ struct scatterlist *sgl_alloc(unsigned long long length, gfp_t gfp, unsigned int *nent_p);void sgl_free_n_order (struct scatterlist *sgl,int nents,int order);void sgl_free_order (struct scatterlist *sgl,int order); +/* Only use sgl_free() when order is 0 */ void sgl_free (struct scatterlist *sgl);#endif /* CONFIG_SGL_ALLOC */
lib/scatterlist.c +8 -6
@@ -554,13 +554,15 @@ EXPORT_SYMBOL(sg_alloc_table_from_pages);#ifdef CONFIG_SGL_ALLOC /** -* sgl_alloc_order - allocate a scatterlist and its pages +* sgl_alloc_order - allocate a scatterlist with equally sized elements * @length: Length in bytes of the scatterlist. Must be at least one -* @order: Second argument for alloc_pages() +* @order: Second argument for alloc_pages(). Each sgl element size will +* be (PAGE_SIZE*2^order) bytes * @chainable: Whether or not to allocate an extra element in the scatterlist -* for scatterlist chaining purposes +* for scatterlist chaining purposes * @gfp: Memory allocation flags -* @nent_p: [out] Number of entries in the scatterlist that have pages +* @nent_p: [out] Number of entries in the scatterlist that have pages. +* Ignored if NULL is given. * * Returns: A pointer to an initialized scatterlist or %NULL upon failure. */
@@ -574,8 +576,8 @@ struct scatterlist *sgl_alloc_order(unsigned long long length,u32 elem_len; nent =round_up (length,PAGE_SIZE << order) >> (PAGE_SHIFT + order); -/* Check for integer overflow */ -if (length > (nent << (PAGE_SHIFT + order))) +/* Integer overflow if: length > nent*2^(PAGE_SHIFT+order) */ +if (ilog2 (length) >ilog2 (nent) +PAGE_SHIFT + order)return NULL ; nalloc = nent;if (chainable) {
Signature · 1 line
2.25.1
Re: [PATCH v5 1/4] sgl_alloc_order: remove 4 GiB limit, sgl_free() warning
Quoted 24 lines · Douglas Gilbert
lib/scatterlist.c +7
-4+++ b/lib/scatterlist.c@@ -554,13 +554,15 @@ EXPORT_SYMBOL(sg_alloc_table_from_pages);#ifdef CONFIG_SGL_ALLOC /** -* sgl_alloc_order - allocate a scatterlist and its pages +* sgl_alloc_order - allocate a scatterlist with equally sized elements * @length: Length in bytes of the scatterlist. Must be at least one -* @order: Second argument for alloc_pages() +* @order: Second argument for alloc_pages(). Each sgl element size will +* be (PAGE_SIZE*2^order) bytes * @chainable: Whether or not to allocate an extra element in the scatterlist -* for scatterlist chaining purposes +* for scatterlist chaining purposes * @gfp: Memory allocation flags -* @nent_p: [out] Number of entries in the scatterlist that have pages +* @nent_p: [out] Number of entries in the scatterlist that have pages. +* Ignored if NULL is given. * * Returns: A pointer to an initialized scatterlist or %NULL upon failure. */
lib/scatterlist.c struct scatterlist *sgl_alloc_order(unsigned long long length, · line 583
Quoted 10 lines
lib/scatterlist.c +2
-2@@ -574,8 +576,8 @@ struct scatterlist *sgl_alloc_order(unsigned long long length,u32 elem_len; nent =round_up (length,PAGE_SIZE << order) >> (PAGE_SHIFT + order); -/* Check for integer overflow */ -if (length > (nent << (PAGE_SHIFT + order))) +/* Integer overflow if: length > nent*2^(PAGE_SHIFT+order) */ +if (ilog2 (length) >ilog2 (nent) +PAGE_SHIFT + order)return NULL ; nalloc = nent;if (chainable) {
This is a little bit too tortured now, how about this:
if (length >> (PAGE_SHIFT + order) >= UINT_MAX)
return NULL;
nent = length >> (PAGE_SHIFT + order);
if (length & ((1ULL << (PAGE_SHIFT + order)) - 1))
nent++;
if (chainable) {
if (check_add_overflow(nent, 1, &nalloc))
return NULL;
}
else
nalloc = nent;Jason
Re: [PATCH v5 1/4] sgl_alloc_order: remove 4 GiB limit, sgl_free() warning
Quoted 38 lines · Jason Gunthorpe
Quoted 24 lines · Douglas Gilbert
lib/scatterlist.c +7
-4+++ b/lib/scatterlist.c@@ -554,13 +554,15 @@ EXPORT_SYMBOL(sg_alloc_table_from_pages);#ifdef CONFIG_SGL_ALLOC /** -* sgl_alloc_order - allocate a scatterlist and its pages +* sgl_alloc_order - allocate a scatterlist with equally sized elements * @length: Length in bytes of the scatterlist. Must be at least one -* @order: Second argument for alloc_pages() +* @order: Second argument for alloc_pages(). Each sgl element size will +* be (PAGE_SIZE*2^order) bytes * @chainable: Whether or not to allocate an extra element in the scatterlist -* for scatterlist chaining purposes +* for scatterlist chaining purposes * @gfp: Memory allocation flags -* @nent_p: [out] Number of entries in the scatterlist that have pages +* @nent_p: [out] Number of entries in the scatterlist that have pages. +* Ignored if NULL is given. * * Returns: A pointer to an initialized scatterlist or %NULL upon failure. */ lib/scatterlist.c struct scatterlist *sgl_alloc_order(unsigned long long length, · line 583
Quoted 10 lines
lib/scatterlist.c +2
-2@@ -574,8 +576,8 @@ struct scatterlist *sgl_alloc_order(unsigned long long length,u32 elem_len; nent =round_up (length,PAGE_SIZE << order) >> (PAGE_SHIFT + order); -/* Check for integer overflow */ -if (length > (nent << (PAGE_SHIFT + order))) +/* Integer overflow if: length > nent*2^(PAGE_SHIFT+order) */ +if (ilog2 (length) >ilog2 (nent) +PAGE_SHIFT + order)return NULL ; nalloc = nent;if (chainable) {This is a little bit too tortured now, how about this:
Quoted 13 lines
if (length >> (PAGE_SHIFT + order) >= UINT_MAX)
return NULL;
nent = length >> (PAGE_SHIFT + order);
if (length & ((1ULL << (PAGE_SHIFT + order)) - 1))
nent++;
if (chainable) {
if (check_add_overflow(nent, 1, &nalloc))
return NULL;
}
else
nalloc = nent;And your proposal is less <<tortured>> ?
I'm looking at performance, not elegance and I'm betting that two ilog2() calls [which boil down to ffs()] are faster than two right-shift-by-n_s and one left-shift-by-n . Perhaps an extra comment could help my code by noting that mathematically:
/* if n > m for positive n and m then: log(n) > log(m) */
My original preference was to drop the check all together but Bart Van Assche (who wrote that function) wanted me to keep it. Any function that takes 'order' (i.e. an exponent) can blow up given a silly value.
The chainable check_add_overflow() call is new and an improvement.
Doug Gilbert
Re: [PATCH v5 1/4] sgl_alloc_order: remove 4 GiB limit, sgl_free() warning
Quoted 51 lines · Douglas Gilbert
Quoted 36 lines · Jason Gunthorpe
Quoted 23 lines · Douglas Gilbert
lib/scatterlist.c +7
-4+++ b/lib/scatterlist.c@@ -554,13 +554,15 @@ EXPORT_SYMBOL(sg_alloc_table_from_pages);#ifdef CONFIG_SGL_ALLOC /** -* sgl_alloc_order - allocate a scatterlist and its pages +* sgl_alloc_order - allocate a scatterlist with equally sized elements * @length: Length in bytes of the scatterlist. Must be at least one -* @order: Second argument for alloc_pages() +* @order: Second argument for alloc_pages(). Each sgl element size will +* be (PAGE_SIZE*2^order) bytes * @chainable: Whether or not to allocate an extra element in the scatterlist -* for scatterlist chaining purposes +* for scatterlist chaining purposes * @gfp: Memory allocation flags -* @nent_p: [out] Number of entries in the scatterlist that have pages +* @nent_p: [out] Number of entries in the scatterlist that have pages. +* Ignored if NULL is given. * * Returns: A pointer to an initialized scatterlist or %NULL upon failure. */ lib/scatterlist.c struct scatterlist *sgl_alloc_order(unsigned long long length, · line 582
Quoted 9 lines
lib/scatterlist.c +2
-2@@ -574,8 +576,8 @@ struct scatterlist *sgl_alloc_order(unsigned long long length,u32 elem_len; nent =round_up (length,PAGE_SIZE << order) >> (PAGE_SHIFT + order); -/* Check for integer overflow */ -if (length > (nent << (PAGE_SHIFT + order))) +/* Integer overflow if: length > nent*2^(PAGE_SHIFT+order) */ +if (ilog2 (length) >ilog2 (nent) +PAGE_SHIFT + order)return NULL ; nalloc = nent;if (chainable) {This is a little bit too tortured now, how about this:
Quoted 13 lines
if (length >> (PAGE_SHIFT + order) >= UINT_MAX) return NULL; nent = length >> (PAGE_SHIFT + order); if (length & ((1ULL << (PAGE_SHIFT + order)) - 1)) nent++; if (chainable) { if (check_add_overflow(nent, 1, &nalloc)) return NULL; } else nalloc = nent;
Quoted 1 line
And your proposal is less <<tortured>> ?
Yes, obviously checking something fits in a variable is less tortured than checking the result of math is correct.
Quoted 5 lines
I'm looking at performance, not elegance and I'm betting that two
ilog2()calls [which boil down toffs()] are faster than two right-shift-by-n_s and one left-shift-by-n . Perhaps an extra comment could help my code by noting that mathematically:
/* if n > m for positive n and m then: log(n) > log(m) */
One instruction difference seems completely irrelavent here.
If you care about micro-optimizing this then please add a check_shr_overflow() just like we have for check_shl_overflow() that has all the right tricks.
Probably:
input_type x = arg >> shift;
if (x != (output_type)x)
fail
return (output_type)x
Is fastest.
Jason
PATCH v5 2/4 scatterlist: add sgl_copy_sgl() function
In reply to Douglas GilbertBoth the SCSI and NVMe subsystems receive user data from the block layer in scatterlist_s (aka scatter gather lists (sgl) which are often arrays). If drivers in those subsystems represent storage (e.g. a ramdisk) or cache "hot" user data then they may also choose to use scatterlist_s. Currently there are no sgl to sgl operations in the kernel. Start with a sgl to sgl copy. Stops when the first of the number of requested bytes to copy, or the source sgl, or the destination sgl is exhausted. So the destination sgl will _not_ grow.
--- include/linux/scatterlist.h | 4 ++ lib/scatterlist.c | 74 +++++++++++++++++++++++++++++++++++++ 2 files changed, 78 insertions(+)
include/linux/scatterlist.h +4 -0
@@ -321,6 +321,10 @@ size_t sg_pcopy_to_buffer(struct scatterlist *sgl, unsigned int nents,size_t sg_zero_buffer (struct scatterlist *sgl,unsigned int nents,size_t buflen,off_t skip); +size_t sgl_copy_sgl (struct scatterlist *d_sgl,unsigned int d_nents,off_t d_skip, +struct scatterlist *s_sgl,unsigned int s_nents,off_t s_skip, +size_t n_bytes); +/* * Maximum number of entries that will be allocated in one piece, if * a list larger than this is required then chaining will be utilized.
lib/scatterlist.c +74 -0
77 lines not on this page · raw message
Signature · 1 line
2.25.1
PATCH v5 3/4 scatterlist: add sgl_compare_sgl() function
In reply to Douglas GilbertAfter enabling copies between scatter gather lists (sgl_s), another storage related operation is to compare two sgl_s. This new function is modelled on NVMe's Compare command and the SCSI VERIFY(BYTCHK=1) command. Like memcmp() this function returns false on the first miscompare and stops comparing.
A helper function called sgl_compare_sgl_idx() is added. It takes an additional parameter (miscompare_idx) which is a pointer. If that pointer is non-NULL and a miscompare is detected (i.e. the function returns false) then the byte index of the first miscompare is written to *miscomapre_idx. Knowing the location of the first miscompare is needed to implement the SCSI COMPARE AND WRITE command properly.
--- include/linux/scatterlist.h | 8 +++ lib/scatterlist.c | 109 ++++++++++++++++++++++++++++++++++++ 2 files changed, 117 insertions(+)
include/linux/scatterlist.h +8 -0
@@ -325,6 +325,14 @@ size_t sgl_copy_sgl(struct scatterlist *d_sgl, unsigned int d_nents, off_t d_ski struct scatterlist *s_sgl, unsigned int s_nents, off_t s_skip, size_t n_bytes); +bool sgl_compare_sgl (struct scatterlist *x_sgl,unsigned int x_nents,off_t x_skip, +struct scatterlist *y_sgl,unsigned int y_nents,off_t y_skip, +size_t n_bytes); + +bool sgl_compare_sgl_idx (struct scatterlist *x_sgl,unsigned int x_nents,off_t x_skip, +struct scatterlist *y_sgl,unsigned int y_nents,off_t y_skip, +size_t n_bytes,size_t *miscompare_idx); +/* * Maximum number of entries that will be allocated in one piece, if * a list larger than this is required then chaining will be utilized.
lib/scatterlist.c +109 -0
112 lines not on this page · raw message
Signature · 1 line
2.25.1
PATCH v5 4/4 scatterlist: add sgl_memset()
In reply to Douglas GilbertThe existing sg_zero_buffer() function is a bit restrictive. For example protection information (PI) blocks are usually initialized to 0xff bytes. As its name suggests sgl_memset() is modelled on memset(). One difference is the type of the val argument which is u8 rather than int. Plus it returns the number of bytes (over)written.
Change implementation of sg_zero_buffer() to call this new function.
--- include/linux/scatterlist.h | 3 ++ lib/scatterlist.c | 65 +++++++++++++++++++++++++------------ 2 files changed, 48 insertions(+), 20 deletions(-)
include/linux/scatterlist.h +3 -0
@@ -333,6 +333,9 @@ bool sgl_compare_sgl_idx(struct scatterlist *x_sgl, unsigned int x_nents, off_t struct scatterlist *y_sgl, unsigned int y_nents, off_t y_skip, size_t n_bytes, size_t *miscompare_idx); +size_t sgl_memset (struct scatterlist *sgl,unsigned int nents,off_t skip, +u8 val,size_t n_bytes); +/* * Maximum number of entries that will be allocated in one piece, if * a list larger than this is required then chaining will be utilized.
lib/scatterlist.c +45 -20
74 lines not on this page · raw message
Signature · 1 line
2.25.1
Re: [PATCH v5 4/4] scatterlist: add sgl_memset()
Quoted 47 lines · Douglas Gilbert
The existing
sg_zero_buffer()function is a bit restrictive. For example protection information (PI) blocks are usually initialized to0xffbytes. As its name suggestssgl_memset()is modelled onmemset(). One difference is the type of the val argument which is u8 rather than int. Plus it returns the number of bytes (over)written.Change implementation of
sg_zero_buffer()to call this new function.Reviewed-by Bodo Stroesser <[email protected]>Signed-off-by Douglas Gilbert <[email protected]>include/linux/scatterlist.h | 3 ++ lib/scatterlist.c | 65 +++++++++++++++++++++++++------------ 2 files changed, 48 insertions(+), 20 deletions(-)include/linux/scatterlist.h +4
-0+++ b/include/linux/scatterlist.h@@ -333,6 +333,9 @@ bool sgl_compare_sgl_idx(struct scatterlist *x_sgl, unsigned int x_nents, off_t struct scatterlist *y_sgl, unsigned int y_nents, off_t y_skip, size_t n_bytes, size_t *miscompare_idx); +size_t sgl_memset (struct scatterlist *sgl,unsigned int nents,off_t skip, +u8 val,size_t n_bytes); +/* * Maximum number of entries that will be allocated in one piece, if * a list larger than this is required then chaining will be utilized. lib/scatterlist.c +1
-13+++ b/lib/scatterlist.c@@ -1038,26 +1038,7 @@ EXPORT_SYMBOL(sg_pcopy_to_buffer);size_t sg_zero_buffer (struct scatterlist *sgl,unsigned int nents,size_t buflen,off_t skip) { -unsigned int offset =0 ; -struct sg_mapping_iter miter; -unsigned int sg_flags =SG_MITER_ATOMIC |SG_MITER_TO_SG ; - -sg_miter_start (&miter, sgl, nents, sg_flags); - -if (!sg_miter_skip (&miter, skip)) -return false; - -while (offset < buflen &&sg_miter_next (&miter)) { -unsigned int len; - - len =min (miter.length, buflen - offset);
lib/scatterlist.c EXPORT_SYMBOL(sg_pcopy_to_buffer); · line 1043
Quoted 10 lines
lib/scatterlist.c +1
-7-memset (miter.addr,0 , len); - - offset += len; - } - -sg_miter_stop (&miter); -return offset; +return sgl_memset (sgl, nents, skip,0 , buflen); }EXPORT_SYMBOL (sg_zero_buffer);
May as well make this one liner a static inline in the header. Just rename this function to sgl_memset so the diff is clearer
Jason
Re: [PATCH v5 4/4] scatterlist: add sgl_memset()
Quoted 53 lines · Jason Gunthorpe
Quoted 52 lines · Douglas Gilbert
The existing
sg_zero_buffer()function is a bit restrictive. For example protection information (PI) blocks are usually initialized to0xffbytes. As its name suggestssgl_memset()is modelled onmemset(). One difference is the type of the val argument which is u8 rather than int. Plus it returns the number of bytes (over)written.Change implementation of
sg_zero_buffer()to call this new function.Reviewed-by Bodo Stroesser <[email protected]>Signed-off-by Douglas Gilbert <[email protected]>include/linux/scatterlist.h | 3 ++ lib/scatterlist.c | 65 +++++++++++++++++++++++++------------ 2 files changed, 48 insertions(+), 20 deletions(-)include/linux/scatterlist.h +4
-0+++ b/include/linux/scatterlist.h@@ -333,6 +333,9 @@ bool sgl_compare_sgl_idx(struct scatterlist *x_sgl, unsigned int x_nents, off_t struct scatterlist *y_sgl, unsigned int y_nents, off_t y_skip, size_t n_bytes, size_t *miscompare_idx); +size_t sgl_memset (struct scatterlist *sgl,unsigned int nents,off_t skip, +u8 val,size_t n_bytes); +/* * Maximum number of entries that will be allocated in one piece, if * a list larger than this is required then chaining will be utilized. lib/scatterlist.c +1
-18+++ b/lib/scatterlist.c@@ -1038,26 +1038,7 @@ EXPORT_SYMBOL(sg_pcopy_to_buffer);size_t sg_zero_buffer (struct scatterlist *sgl,unsigned int nents,size_t buflen,off_t skip) { -unsigned int offset =0 ; -struct sg_mapping_iter miter; -unsigned int sg_flags =SG_MITER_ATOMIC |SG_MITER_TO_SG ; - -sg_miter_start (&miter, sgl, nents, sg_flags); - -if (!sg_miter_skip (&miter, skip)) -return false; - -while (offset < buflen &&sg_miter_next (&miter)) { -unsigned int len; - - len =min (miter.length, buflen - offset); -memset (miter.addr,0 , len); - - offset += len; - } -
Quoted 8 lines
lib/scatterlist.c EXPORT_SYMBOL(sg_pcopy_to_buffer); · line 1043
Quoted 5 lines
lib/scatterlist.c +1
-2-sg_miter_stop (&miter); -return offset; +return sgl_memset (sgl, nents, skip,0 , buflen); }EXPORT_SYMBOL (sg_zero_buffer);May as well make this one liner a static inline in the header. Just rename this function to
sgl_memsetso the diff is clearer
Yes, fine. I can roll a new version.
Doug Gilbert