Message ID | 20231203192422.539300-1-yury.norov@gmail.com |
---|---|
Headers | show |
Series | bitops: add atomic find_bit() operations | expand |
On Sun, Dec 03, 2023 at 11:23:47AM -0800, Yury Norov wrote: > Add helpers around test_and_{set,clear}_bit() that allow to search for > clear or set bits and flip them atomically. > > The target patterns may look like this: > > for (idx = 0; idx < nbits; idx++) > if (test_and_clear_bit(idx, bitmap)) > do_something(idx); > > Or like this: > > do { > bit = find_first_bit(bitmap, nbits); > if (bit >= nbits) > return nbits; > } while (!test_and_clear_bit(bit, bitmap)); > return bit; > > In both cases, the opencoded loop may be converted to a single function > or iterator call. Correspondingly: > > for_each_test_and_clear_bit(idx, bitmap, nbits) > do_something(idx); > > Or: > return find_and_clear_bit(bitmap, nbits); > > Obviously, the less routine code people have to write themself, the > less probability to make a mistake. > > Those are not only handy helpers but also resolve a non-trivial > issue of using non-atomic find_bit() together with atomic > test_and_{set,clear)_bit(). > > The trick is that find_bit() implies that the bitmap is a regular > non-volatile piece of memory, and compiler is allowed to use such > optimization techniques like re-fetching memory instead of caching it. > > For example, find_first_bit() is implemented like this: > > for (idx = 0; idx * BITS_PER_LONG < sz; idx++) { > val = addr[idx]; > if (val) { > sz = min(idx * BITS_PER_LONG + __ffs(val), sz); > break; > } > } > > On register-memory architectures, like x86, compiler may decide to > access memory twice - first time to compare against 0, and second time > to fetch its value to pass it to __ffs(). > > When running find_first_bit() on volatile memory, the memory may get > changed in-between, and for instance, it may lead to passing 0 to > __ffs(), which is undefined. This is a potentially dangerous call. > > find_and_clear_bit() as a wrapper around test_and_clear_bit() > naturally treats underlying bitmap as a volatile memory and prevents > compiler from such optimizations. > > Now that KCSAN is catching exactly this type of situations and warns on > undercover memory modifications. We can use it to reveal improper usage > of find_bit(), and convert it to atomic find_and_*_bit() as appropriate. > > The 1st patch of the series adds the following atomic primitives: > > find_and_set_bit(addr, nbits); > find_and_set_next_bit(addr, nbits, start); > ... > > Here find_and_{set,clear} part refers to the corresponding > test_and_{set,clear}_bit function. Suffixes like _wrap or _lock > derive their semantics from corresponding find() or test() functions. > > For brevity, the naming omits the fact that we search for zero bit in > find_and_set, and correspondingly search for set bit in find_and_clear > functions. > > The patch also adds iterators with atomic semantics, like > for_each_test_and_set_bit(). Here, the naming rule is to simply prefix > corresponding atomic operation with 'for_each'. > > This series is a result of discussion [1]. All find_bit() functions imply > exclusive access to the bitmaps. However, KCSAN reports quite a number > of warnings related to find_bit() API. Some of them are not pointing > to real bugs because in many situations people intentionally allow > concurrent bitmap operations. > > If so, find_bit() can be annotated such that KCSAN will ignore it: > > bit = data_race(find_first_bit(bitmap, nbits)); > > This series addresses the other important case where people really need > atomic find ops. As the following patches show, the resulting code > looks safer and more verbose comparing to opencoded loops followed by > atomic bit flips. > > In [1] Mirsad reported 2% slowdown in a single-thread search test when > switching find_bit() function to treat bitmaps as volatile arrays. On > the other hand, kernel robot in the same thread reported +3.7% to the > performance of will-it-scale.per_thread_ops test. > > Assuming that our compilers are sane and generate better code against > properly annotated data, the above discrepancy doesn't look weird. When > running on non-volatile bitmaps, plain find_bit() outperforms atomic > find_and_bit(), and vice-versa. ... In some cases the better improvements can be achieved by switching the (very) old code to utilise IDA framework.
Hello Yury! On Sun 03-12-23 11:23:47, Yury Norov wrote: > Add helpers around test_and_{set,clear}_bit() that allow to search for > clear or set bits and flip them atomically. > > The target patterns may look like this: > > for (idx = 0; idx < nbits; idx++) > if (test_and_clear_bit(idx, bitmap)) > do_something(idx); > > Or like this: > > do { > bit = find_first_bit(bitmap, nbits); > if (bit >= nbits) > return nbits; > } while (!test_and_clear_bit(bit, bitmap)); > return bit; > > In both cases, the opencoded loop may be converted to a single function > or iterator call. Correspondingly: > > for_each_test_and_clear_bit(idx, bitmap, nbits) > do_something(idx); > > Or: > return find_and_clear_bit(bitmap, nbits); These are fine cleanups but they actually don't address the case that has triggered all these changes - namely the xarray use of find_next_bit() in xas_find_chunk(). ... > This series is a result of discussion [1]. All find_bit() functions imply > exclusive access to the bitmaps. However, KCSAN reports quite a number > of warnings related to find_bit() API. Some of them are not pointing > to real bugs because in many situations people intentionally allow > concurrent bitmap operations. > > If so, find_bit() can be annotated such that KCSAN will ignore it: > > bit = data_race(find_first_bit(bitmap, nbits)); No, this is not a correct thing to do. If concurrent bitmap changes can happen, find_first_bit() as it is currently implemented isn't ever a safe choice because it can call __ffs(0) which is dangerous as you properly note above. I proposed adding READ_ONCE() into find_first_bit() / find_next_bit() implementation to fix this issue but you disliked that. So other option we have is adding find_first_bit() and find_next_bit() variants that take volatile 'addr' and we have to use these in code like xas_find_chunk() which cannot be converted to your new helpers. > This series addresses the other important case where people really need > atomic find ops. As the following patches show, the resulting code > looks safer and more verbose comparing to opencoded loops followed by > atomic bit flips. > > In [1] Mirsad reported 2% slowdown in a single-thread search test when > switching find_bit() function to treat bitmaps as volatile arrays. On > the other hand, kernel robot in the same thread reported +3.7% to the > performance of will-it-scale.per_thread_ops test. It was actually me who reported the regression here [2] but whatever :) [2] https://lore.kernel.org/all/20231011150252.32737-1-jack@suse.cz > Assuming that our compilers are sane and generate better code against > properly annotated data, the above discrepancy doesn't look weird. When > running on non-volatile bitmaps, plain find_bit() outperforms atomic > find_and_bit(), and vice-versa. > > So, all users of find_bit() API, where heavy concurrency is expected, > are encouraged to switch to atomic find_and_bit() as appropriate. Well, all users where any concurrency can happen should switch. Otherwise they are prone to the (admittedly mostly theoretical) data race issue. Honza
On Mon, Dec 04, 2023 at 07:51:01PM +0100, Jan Kara wrote: > Hello Yury! > > On Sun 03-12-23 11:23:47, Yury Norov wrote: > > Add helpers around test_and_{set,clear}_bit() that allow to search for > > clear or set bits and flip them atomically. > > > > The target patterns may look like this: > > > > for (idx = 0; idx < nbits; idx++) > > if (test_and_clear_bit(idx, bitmap)) > > do_something(idx); > > > > Or like this: > > > > do { > > bit = find_first_bit(bitmap, nbits); > > if (bit >= nbits) > > return nbits; > > } while (!test_and_clear_bit(bit, bitmap)); > > return bit; > > > > In both cases, the opencoded loop may be converted to a single function > > or iterator call. Correspondingly: > > > > for_each_test_and_clear_bit(idx, bitmap, nbits) > > do_something(idx); > > > > Or: > > return find_and_clear_bit(bitmap, nbits); > > These are fine cleanups but they actually don't address the case that has > triggered all these changes - namely the xarray use of find_next_bit() in > xas_find_chunk(). > > ... > > This series is a result of discussion [1]. All find_bit() functions imply > > exclusive access to the bitmaps. However, KCSAN reports quite a number > > of warnings related to find_bit() API. Some of them are not pointing > > to real bugs because in many situations people intentionally allow > > concurrent bitmap operations. > > > > If so, find_bit() can be annotated such that KCSAN will ignore it: > > > > bit = data_race(find_first_bit(bitmap, nbits)); > > No, this is not a correct thing to do. If concurrent bitmap changes can > happen, find_first_bit() as it is currently implemented isn't ever a safe > choice because it can call __ffs(0) which is dangerous as you properly note > above. I proposed adding READ_ONCE() into find_first_bit() / find_next_bit() > implementation to fix this issue but you disliked that. So other option we > have is adding find_first_bit() and find_next_bit() variants that take > volatile 'addr' and we have to use these in code like xas_find_chunk() > which cannot be converted to your new helpers. Here is some examples when concurrent operations with plain find_bit() are acceptable: - two threads running find_*_bit(): safe wrt ffs(0) and returns correct value, because underlying bitmap is unchanged; - find_next_bit() in parallel with set or clear_bit(), when modifying a bit prior to the start bit to search: safe and correct; - find_first_bit() in parallel with set_bit(): safe, but may return wrong bit number; - find_first_zero_bit() in parallel with clear_bit(): same as above. In last 2 cases find_bit() may not return a correct bit number, but it may be OK if caller requires any (not exactly first) set or clear bit, correspondingly. In such cases, KCSAN may be safely silenced. > > This series addresses the other important case where people really need > > atomic find ops. As the following patches show, the resulting code > > looks safer and more verbose comparing to opencoded loops followed by > > atomic bit flips. > > > > In [1] Mirsad reported 2% slowdown in a single-thread search test when > > switching find_bit() function to treat bitmaps as volatile arrays. On > > the other hand, kernel robot in the same thread reported +3.7% to the > > performance of will-it-scale.per_thread_ops test. > > It was actually me who reported the regression here [2] but whatever :) > > [2] https://lore.kernel.org/all/20231011150252.32737-1-jack@suse.cz My apologize. > > Assuming that our compilers are sane and generate better code against > > properly annotated data, the above discrepancy doesn't look weird. When > > running on non-volatile bitmaps, plain find_bit() outperforms atomic > > find_and_bit(), and vice-versa. > > > > So, all users of find_bit() API, where heavy concurrency is expected, > > are encouraged to switch to atomic find_and_bit() as appropriate. > > Well, all users where any concurrency can happen should switch. Otherwise > they are prone to the (admittedly mostly theoretical) data race issue. > > Honza > -- > Jan Kara <jack@suse.com> > SUSE Labs, CR
On Tue 05-12-23 21:22:59, Yury Norov wrote: > On Mon, Dec 04, 2023 at 07:51:01PM +0100, Jan Kara wrote: > > > This series is a result of discussion [1]. All find_bit() functions imply > > > exclusive access to the bitmaps. However, KCSAN reports quite a number > > > of warnings related to find_bit() API. Some of them are not pointing > > > to real bugs because in many situations people intentionally allow > > > concurrent bitmap operations. > > > > > > If so, find_bit() can be annotated such that KCSAN will ignore it: > > > > > > bit = data_race(find_first_bit(bitmap, nbits)); > > > > No, this is not a correct thing to do. If concurrent bitmap changes can > > happen, find_first_bit() as it is currently implemented isn't ever a safe > > choice because it can call __ffs(0) which is dangerous as you properly note > > above. I proposed adding READ_ONCE() into find_first_bit() / find_next_bit() > > implementation to fix this issue but you disliked that. So other option we > > have is adding find_first_bit() and find_next_bit() variants that take > > volatile 'addr' and we have to use these in code like xas_find_chunk() > > which cannot be converted to your new helpers. > > Here is some examples when concurrent operations with plain find_bit() > are acceptable: > > - two threads running find_*_bit(): safe wrt ffs(0) and returns correct > value, because underlying bitmap is unchanged; > - find_next_bit() in parallel with set or clear_bit(), when modifying > a bit prior to the start bit to search: safe and correct; > - find_first_bit() in parallel with set_bit(): safe, but may return wrong > bit number; > - find_first_zero_bit() in parallel with clear_bit(): same as above. > > In last 2 cases find_bit() may not return a correct bit number, but > it may be OK if caller requires any (not exactly first) set or clear > bit, correspondingly. > > In such cases, KCSAN may be safely silenced. True - but these are special cases. In particular the case in xas_find_chunk() is not any of these special cases. It is using find_next_bit() which is can be racing with clear_bit(). So what are your plans for such usecase? Honza