diff --git a/Documentation/memory-barriers.txt b/Documentation/memory-barriers.txt index c61d8b876fdb..4710845dbac4 100644 --- a/Documentation/memory-barriers.txt +++ b/Documentation/memory-barriers.txt @@ -19,6 +19,7 @@ Contents: - Control dependencies. - SMP barrier pairing. - Examples of memory barrier sequences. + - Read memory barriers vs load speculation. (*) Explicit kernel barriers. @@ -248,7 +249,7 @@ And there are a number of things that _must_ or _must_not_ be assumed: we may get either of: STORE *A = X; Y = LOAD *A; - STORE *A = Y; + STORE *A = Y = X; ========================= @@ -344,9 +345,12 @@ Memory barriers come in four basic varieties: (4) General memory barriers. - A general memory barrier is a combination of both a read memory barrier - and a write memory barrier. It is a partial ordering over both loads and - stores. + A general memory barrier gives a guarantee that all the LOAD and STORE + operations specified before the barrier will appear to happen before all + the LOAD and STORE operations specified after the barrier with respect to + the other components of the system. + + A general memory barrier is a partial ordering over both loads and stores. General memory barriers imply both read and write memory barriers, and so can substitute for either. @@ -546,9 +550,9 @@ write barrier, though, again, a general barrier is viable: =============== =============== a = 1; - b = 2; x = a; + b = 2; x = b; - y = b; + y = a; Or: @@ -563,6 +567,18 @@ Or: Basically, the read barrier always has to be there, even though it can be of the "weaker" type. +[!] Note that the stores before the write barrier would normally be expected to +match the loads after the read barrier or data dependency barrier, and vice +versa: + + CPU 1 CPU 2 + =============== =============== + a = 1; }---- --->{ v = c + b = 2; } \ / { w = d + \ + c = 3; } / \ { x = a; + d = 4; }---- --->{ y = b; + EXAMPLES OF MEMORY BARRIER SEQUENCES ------------------------------------ @@ -600,8 +616,8 @@ STORE B, STORE C } all occuring before the unordered set of { STORE D, STORE E | | +------+ +-------+ : : | - | Sequence in which stores committed to memory system - | by CPU 1 + | Sequence in which stores are committed to the + | memory system by CPU 1 V @@ -683,14 +699,12 @@ then the following will occur: | : : | | | : : | CPU 2 | | +-------+ | | - \ | X->9 |------>| | - \ +-------+ | | - ----->| B->2 | | | - +-------+ | | - Makes sure all effects ---> ddddddddddddddddd | | - prior to the store of C +-------+ | | - are perceptible to | B->2 |------>| | - successive loads +-------+ | | + | | X->9 |------>| | + | +-------+ | | + Makes sure all effects ---> \ ddddddddddddddddd | | + prior to the store of C \ +-------+ | | + are perceptible to ----->| B->2 |------>| | + subsequent loads +-------+ | | : : +-------+ @@ -699,73 +713,239 @@ following sequence of events: CPU 1 CPU 2 ======================= ======================= + { A = 0, B = 9 } STORE A=1 - STORE B=2 - STORE C=3 - STORE D=4 - STORE E=5 - LOAD A + STORE B=2 LOAD B - LOAD C - LOAD D - LOAD E + LOAD A Without intervention, CPU 2 may then choose to perceive the events on CPU 1 in some effectively random order, despite the write barrier issued by CPU 1: - +-------+ : : - | | +------+ - | |------>| C=3 | } - | | : +------+ } - | | : | A=1 | } - | | : +------+ } - | CPU 1 | : | B=2 | }--- - | | +------+ } \ - | | wwwwwwwwwwwww} \ - | | +------+ } \ : : +-------+ - | | : | E=5 | } \ +-------+ | | - | | : +------+ } \ { | C->3 |------>| | - | |------>| D=4 | } \ { +-------+ : | | - | | +------+ \ { | E->5 | : | | - +-------+ : : \ { +-------+ : | | - Transfer -->{ | A->1 | : | CPU 2 | - from CPU 1 { +-------+ : | | - to CPU 2 { | D->4 | : | | - { +-------+ : | | - { | B->2 |------>| | - +-------+ | | - : : +-------+ + +-------+ : : : : + | | +------+ +-------+ + | |------>| A=1 |------ --->| A->0 | + | | +------+ \ +-------+ + | CPU 1 | wwwwwwwwwwwwwwww \ --->| B->9 | + | | +------+ | +-------+ + | |------>| B=2 |--- | : : + | | +------+ \ | : : +-------+ + +-------+ : : \ | +-------+ | | + ---------->| B->2 |------>| | + | +-------+ | CPU 2 | + | | A->0 |------>| | + | +-------+ | | + | : : +-------+ + \ : : + \ +-------+ + ---->| A->1 | + +-------+ + : : -If, however, a read barrier were to be placed between the load of C and the -load of D on CPU 2, then the partial ordering imposed by CPU 1 will be -perceived correctly by CPU 2. +If, however, a read barrier were to be placed between the load of E and the +load of A on CPU 2: - +-------+ : : - | | +------+ - | |------>| C=3 | } - | | : +------+ } - | | : | A=1 | }--- - | | : +------+ } \ - | CPU 1 | : | B=2 | } \ - | | +------+ \ - | | wwwwwwwwwwwwwwww \ - | | +------+ \ : : +-------+ - | | : | E=5 | } \ +-------+ | | - | | : +------+ }--- \ { | C->3 |------>| | - | |------>| D=4 | } \ \ { +-------+ : | | - | | +------+ \ -->{ | B->2 | : | | - +-------+ : : \ { +-------+ : | | - \ { | A->1 | : | CPU 2 | - \ +-------+ | | - At this point the read ----> \ rrrrrrrrrrrrrrrrr | | - barrier causes all effects \ +-------+ | | - prior to the storage of C \ { | E->5 | : | | - to be perceptible to CPU 2 -->{ +-------+ : | | - { | D->4 |------>| | - +-------+ | | - : : +-------+ + CPU 1 CPU 2 + ======================= ======================= + { A = 0, B = 9 } + STORE A=1 + + STORE B=2 + LOAD B + + LOAD A + +then the partial ordering imposed by CPU 1 will be perceived correctly by CPU +2: + + +-------+ : : : : + | | +------+ +-------+ + | |------>| A=1 |------ --->| A->0 | + | | +------+ \ +-------+ + | CPU 1 | wwwwwwwwwwwwwwww \ --->| B->9 | + | | +------+ | +-------+ + | |------>| B=2 |--- | : : + | | +------+ \ | : : +-------+ + +-------+ : : \ | +-------+ | | + ---------->| B->2 |------>| | + | +-------+ | CPU 2 | + | : : | | + | : : | | + At this point the read ----> \ rrrrrrrrrrrrrrrrr | | + barrier causes all effects \ +-------+ | | + prior to the storage of B ---->| A->1 |------>| | + to be perceptible to CPU 2 +-------+ | | + : : +-------+ + + +To illustrate this more completely, consider what could happen if the code +contained a load of A either side of the read barrier: + + CPU 1 CPU 2 + ======================= ======================= + { A = 0, B = 9 } + STORE A=1 + + STORE B=2 + LOAD B + LOAD A [first load of A] + + LOAD A [second load of A] + +Even though the two loads of A both occur after the load of B, they may both +come up with different values: + + +-------+ : : : : + | | +------+ +-------+ + | |------>| A=1 |------ --->| A->0 | + | | +------+ \ +-------+ + | CPU 1 | wwwwwwwwwwwwwwww \ --->| B->9 | + | | +------+ | +-------+ + | |------>| B=2 |--- | : : + | | +------+ \ | : : +-------+ + +-------+ : : \ | +-------+ | | + ---------->| B->2 |------>| | + | +-------+ | CPU 2 | + | : : | | + | : : | | + | +-------+ | | + | | A->0 |------>| 1st | + | +-------+ | | + At this point the read ----> \ rrrrrrrrrrrrrrrrr | | + barrier causes all effects \ +-------+ | | + prior to the storage of B ---->| A->1 |------>| 2nd | + to be perceptible to CPU 2 +-------+ | | + : : +-------+ + + +But it may be that the update to A from CPU 1 becomes perceptible to CPU 2 +before the read barrier completes anyway: + + +-------+ : : : : + | | +------+ +-------+ + | |------>| A=1 |------ --->| A->0 | + | | +------+ \ +-------+ + | CPU 1 | wwwwwwwwwwwwwwww \ --->| B->9 | + | | +------+ | +-------+ + | |------>| B=2 |--- | : : + | | +------+ \ | : : +-------+ + +-------+ : : \ | +-------+ | | + ---------->| B->2 |------>| | + | +-------+ | CPU 2 | + | : : | | + \ : : | | + \ +-------+ | | + ---->| A->1 |------>| 1st | + +-------+ | | + rrrrrrrrrrrrrrrrr | | + +-------+ | | + | A->1 |------>| 2nd | + +-------+ | | + : : +-------+ + + +The guarantee is that the second load will always come up with A == 1 if the +load of B came up with B == 2. No such guarantee exists for the first load of +A; that may come up with either A == 0 or A == 1. + + +READ MEMORY BARRIERS VS LOAD SPECULATION +---------------------------------------- + +Many CPUs speculate with loads: that is they see that they will need to load an +item from memory, and they find a time where they're not using the bus for any +other loads, and so do the load in advance - even though they haven't actually +got to that point in the instruction execution flow yet. This permits the +actual load instruction to potentially complete immediately because the CPU +already has the value to hand. + +It may turn out that the CPU didn't actually need the value - perhaps because a +branch circumvented the load - in which case it can discard the value or just +cache it for later use. + +Consider: + + CPU 1 CPU 2 + ======================= ======================= + LOAD B + DIVIDE } Divide instructions generally + DIVIDE } take a long time to perform + LOAD A + +Which might appear as this: + + : : +-------+ + +-------+ | | + --->| B->2 |------>| | + +-------+ | CPU 2 | + : :DIVIDE | | + +-------+ | | + The CPU being busy doing a ---> --->| A->0 |~~~~ | | + division speculates on the +-------+ ~ | | + LOAD of A : : ~ | | + : :DIVIDE | | + : : ~ | | + Once the divisions are complete --> : : ~-->| | + the CPU can then perform the : : | | + LOAD with immediate effect : : +-------+ + + +Placing a read barrier or a data dependency barrier just before the second +load: + + CPU 1 CPU 2 + ======================= ======================= + LOAD B + DIVIDE + DIVIDE + + LOAD A + +will force any value speculatively obtained to be reconsidered to an extent +dependent on the type of barrier used. If there was no change made to the +speculated memory location, then the speculated value will just be used: + + : : +-------+ + +-------+ | | + --->| B->2 |------>| | + +-------+ | CPU 2 | + : :DIVIDE | | + +-------+ | | + The CPU being busy doing a ---> --->| A->0 |~~~~ | | + division speculates on the +-------+ ~ | | + LOAD of A : : ~ | | + : :DIVIDE | | + : : ~ | | + : : ~ | | + rrrrrrrrrrrrrrrr~ | | + : : ~ | | + : : ~-->| | + : : | | + : : +-------+ + + +but if there was an update or an invalidation from another CPU pending, then +the speculation will be cancelled and the value reloaded: + + : : +-------+ + +-------+ | | + --->| B->2 |------>| | + +-------+ | CPU 2 | + : :DIVIDE | | + +-------+ | | + The CPU being busy doing a ---> --->| A->0 |~~~~ | | + division speculates on the +-------+ ~ | | + LOAD of A : : ~ | | + : :DIVIDE | | + : : ~ | | + : : ~ | | + rrrrrrrrrrrrrrrrr | | + +-------+ | | + The speculation is discarded ---> --->| A->1 |------>| | + and an updated value is +-------+ | | + retrieved : : +-------+ ======================== @@ -901,7 +1081,7 @@ IMPLICIT KERNEL MEMORY BARRIERS =============================== Some of the other functions in the linux kernel imply memory barriers, amongst -which are locking, scheduling and memory allocation functions. +which are locking and scheduling functions. This specification is a _minimum_ guarantee; any particular architecture may provide more substantial guarantees, but these may not be relied upon outside @@ -966,6 +1146,20 @@ equivalent to a full barrier, but a LOCK followed by an UNLOCK is not. barriers is that the effects instructions outside of a critical section may seep into the inside of the critical section. +A LOCK followed by an UNLOCK may not be assumed to be full memory barrier +because it is possible for an access preceding the LOCK to happen after the +LOCK, and an access following the UNLOCK to happen before the UNLOCK, and the +two accesses can themselves then cross: + + *A = a; + LOCK + UNLOCK + *B = b; + +may occur as: + + LOCK, STORE *B, STORE *A, UNLOCK + Locks and semaphores may not provide any guarantee of ordering on UP compiled systems, and so cannot be counted on in such a situation to actually achieve anything at all - especially with respect to I/O accesses - unless combined @@ -1016,8 +1210,6 @@ Other functions that imply barriers: (*) schedule() and similar imply full memory barriers. - (*) Memory allocation and release functions imply full memory barriers. - ================================= INTER-CPU LOCKING BARRIER EFFECTS