Re: [PATCH v18 05/10] xbitmap: add more operations
On Thu, Dec 07, 2017 at 08:01:24PM +0800, Wei Wang wrote: > On 12/03/2017 09:50 AM, Tetsuo Handa wrote: > > Matthew Wilcox wrote: > > > On Fri, Dec 01, 2017 at 03:09:08PM +, Wang, Wei W wrote: > > > > On Friday, December 1, 2017 9:02 PM, Tetsuo Handa wrote: > > > > > If start == end is legal, > > > > > > > > > > for (; start < end; start = (start | (IDA_BITMAP_BITS - 1)) + 1) { > > > > > > > > > > makes this loop do nothing because 10 < 10 is false. > > > > How about "start <= end "? > > > Don't ask Tetsuo for his opinion, write some userspace code that uses it. > > > > > Please be sure to prepare for "end == -1UL" case, for "start < end" will > > become > > true when "start = (start | (IDA_BITMAP_BITS - 1)) + 1" made "start == 0" > > due to > > overflow. > > I think there is one more corner case with this API: searching for bit "1" > from [0, ULONG_MAX] while no bit is set in the range, there appear to be no > possible value that we can return (returning "end + 1" will be "ULONG_MAX + > 1", which is 0) > I plan to make the "end" be exclusive of the searching, that is, [start, > end), and return "end" if no such bit is found. > > For cases like [16, 16), returning 16 doesn't mean bit 16 is 1 or 0, it > simply means there is no bits to search in the given range, since 16 is > exclusive. > > Please let me know if you have a different thought. > > Best, > Wei Matthew is right though - you want to include tests for all these corner cases. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v18 05/10] xbitmap: add more operations
On 12/03/2017 09:50 AM, Tetsuo Handa wrote: Matthew Wilcox wrote: On Fri, Dec 01, 2017 at 03:09:08PM +, Wang, Wei W wrote: On Friday, December 1, 2017 9:02 PM, Tetsuo Handa wrote: If start == end is legal, for (; start < end; start = (start | (IDA_BITMAP_BITS - 1)) + 1) { makes this loop do nothing because 10 < 10 is false. How about "start <= end "? Don't ask Tetsuo for his opinion, write some userspace code that uses it. Please be sure to prepare for "end == -1UL" case, for "start < end" will become true when "start = (start | (IDA_BITMAP_BITS - 1)) + 1" made "start == 0" due to overflow. I think there is one more corner case with this API: searching for bit "1" from [0, ULONG_MAX] while no bit is set in the range, there appear to be no possible value that we can return (returning "end + 1" will be "ULONG_MAX + 1", which is 0) I plan to make the "end" be exclusive of the searching, that is, [start, end), and return "end" if no such bit is found. For cases like [16, 16), returning 16 doesn't mean bit 16 is 1 or 0, it simply means there is no bits to search in the given range, since 16 is exclusive. Please let me know if you have a different thought. Best, Wei ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v18 05/10] xbitmap: add more operations
On Fri, Dec 01, 2017 at 03:09:08PM +, Wang, Wei W wrote: > On Friday, December 1, 2017 9:02 PM, Tetsuo Handa wrote: > > If start == end is legal, > > > >for (; start < end; start = (start | (IDA_BITMAP_BITS - 1)) + 1) { > > > > makes this loop do nothing because 10 < 10 is false. > > How about "start <= end "? Don't ask Tetsuo for his opinion, write some userspace code that uses it. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
RE: [PATCH v18 05/10] xbitmap: add more operations
On Friday, December 1, 2017 9:02 PM, Tetsuo Handa wrote: > Wei Wang wrote: > > On 11/30/2017 06:34 PM, Tetsuo Handa wrote: > > > Wei Wang wrote: > > >> + * @start: the start of the bit range, inclusive > > >> + * @end: the end of the bit range, inclusive > > >> + * > > >> + * This function is used to clear a bit in the xbitmap. If all the > > >> +bits of the > > >> + * bitmap are 0, the bitmap will be freed. > > >> + */ > > >> +void xb_clear_bit_range(struct xb *xb, unsigned long start, > > >> +unsigned long end) { > > >> +struct radix_tree_root *root = >xbrt; > > >> +struct radix_tree_node *node; > > >> +void **slot; > > >> +struct ida_bitmap *bitmap; > > >> +unsigned int nbits; > > >> + > > >> +for (; start < end; start = (start | (IDA_BITMAP_BITS - 1)) + > > >> 1) { > > >> +unsigned long index = start / IDA_BITMAP_BITS; > > >> +unsigned long bit = start % IDA_BITMAP_BITS; > > >> + > > >> +bitmap = __radix_tree_lookup(root, index, , ); > > >> +if (radix_tree_exception(bitmap)) { > > >> +unsigned long ebit = bit + 2; > > >> +unsigned long tmp = (unsigned long)bitmap; > > >> + > > >> +nbits = min(end - start + 1, BITS_PER_LONG - > > >> ebit); > > > "nbits = min(end - start + 1," seems to expect that start == end is > > > legal for clearing only 1 bit. But this function is no-op if start == end. > > > Please clarify what "inclusive" intended. > > > > If xb_clear_bit_range(xb,10,10), then it is effectively the same as > > xb_clear_bit(10). Why would it be illegal? > > > > "@start inclusive" means that the @start will also be included to be > > cleared. > > If start == end is legal, > >for (; start < end; start = (start | (IDA_BITMAP_BITS - 1)) + 1) { > > makes this loop do nothing because 10 < 10 is false. How about "start <= end "? Best, Wei ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v18 05/10] xbitmap: add more operations
On Fri, Dec 01, 2017 at 10:02:01PM +0900, Tetsuo Handa wrote: > If start == end is legal, > >for (; start < end; start = (start | (IDA_BITMAP_BITS - 1)) + 1) { > > makes this loop do nothing because 10 < 10 is false. ... and this is why we add tests to the test-suite! ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v18 05/10] xbitmap: add more operations
On 11/30/2017 06:34 PM, Tetsuo Handa wrote: Wei Wang wrote: + * @start: the start of the bit range, inclusive + * @end: the end of the bit range, inclusive + * + * This function is used to clear a bit in the xbitmap. If all the bits of the + * bitmap are 0, the bitmap will be freed. + */ +void xb_clear_bit_range(struct xb *xb, unsigned long start, unsigned long end) +{ + struct radix_tree_root *root = >xbrt; + struct radix_tree_node *node; + void **slot; + struct ida_bitmap *bitmap; + unsigned int nbits; + + for (; start < end; start = (start | (IDA_BITMAP_BITS - 1)) + 1) { + unsigned long index = start / IDA_BITMAP_BITS; + unsigned long bit = start % IDA_BITMAP_BITS; + + bitmap = __radix_tree_lookup(root, index, , ); + if (radix_tree_exception(bitmap)) { + unsigned long ebit = bit + 2; + unsigned long tmp = (unsigned long)bitmap; + + nbits = min(end - start + 1, BITS_PER_LONG - ebit); "nbits = min(end - start + 1," seems to expect that start == end is legal for clearing only 1 bit. But this function is no-op if start == end. Please clarify what "inclusive" intended. If xb_clear_bit_range(xb,10,10), then it is effectively the same as xb_clear_bit(10). Why would it be illegal? "@start inclusive" means that the @start will also be included to be cleared. +static inline __always_inline void bitmap_clear(unsigned long *map, + unsigned int start, + unsigned int nbits) +{ + if (__builtin_constant_p(nbits) && nbits == 1) + __clear_bit(start, map); + else if (__builtin_constant_p(start & 7) && IS_ALIGNED(start, 8) && +__builtin_constant_p(nbits & 7) && IS_ALIGNED(nbits, 8)) It looks strange to apply __builtin_constant_p test to variables after "& 7". I think this is normal - if the variables are known at compile time, the calculation will be done at compile time (termed constant folding). Best, Wei ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v18 05/10] xbitmap: add more operations
On Thu, Nov 30, 2017 at 10:35:03PM +0900, Tetsuo Handa wrote: > According to xb_set_bit(), it seems to me that we are trying to avoid memory > allocation > for "struct ida_bitmap" when all set bits within a 1024-bits bitmap reside in > the first > 61 bits. > > But does such saving help? Is there characteristic bias that majority of set > bits resides > in the first 61 bits, for "bit" is "unsigned long" which holds a page number > (isn't it)? > If no such bias, wouldn't eliminating radix_tree_exception() case and always > storing > "struct ida_bitmap" simplifies the code (and make the processing faster)? It happens all the time. The vast majority of users of the IDA set low bits. Also, it's the first 62 bits -- going up to 63 bits with the XArray rewrite. I do plan to redo the xbitmap on top of the XArray; I'm just trying to get the XArray merged first. The IDA and xbitmap code will share much more code when that happens. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization