Alexander Terekhov wrote:
Peter Dimov wrote: [...]
Are you sure that keeping a reservation alive for extended periods of time does not incur a performance penalty, BTW?
I've never heard that it can.
Could this be the reason for the preliminary technical note?
I'm also puzzled.
Here's a snippet from the official "Programming Environments Manual" referenced by that puzzling preliminary notice itself. It explicitly mentions dangling lwarxs, says that it is good for forward progress, and shows yet another dangling lwarx in "better performance may be obtained" illustration. <quote> E.1 General Information The following points provide general information about the lwarx and stwcx. instructions: - It is acceptable to execute an lwarx instruction for which no stwcx. instruction is executed. Such a dangling lwarx instruction occurs in the example shown in Section E.2.5 , "Test and Set," if the value loaded is not zero. - To increase the likelihood that forward progress is made, it is important that looping on lwarx/stwcx. pairs be minimized. For example, in the sequence shown in Section E.2.5 , "Test and Set," this is achieved by testing the old value before attempting the store -- were the order reversed, more stwcx. instructions might be executed, and reservations might more often be lost between the lwarx and the stwcx. instructions. - The manner in which lwarx and stwcx. are communicated to other processors and mechanisms, and between levels of the memory subsystem within a given processor, is implementation-dependent. In some implementations, performance may be improved by minimizing looping on an lwarx instruction that fails to return a desired value. For example, in the example provided in Section E.2.5 , "Test and Set," if the program stays in the loop until the word loaded is zero, the programmer can change the "bne- $+12" to "bne- loop." In some implementations, better performance may be obtained by using an ordinary load instruction to do the initial checking of the value, as follows: loop: lwz r5,0(r3) #load the word cmpwi r5,0 #loop back if word bne- loop #not equal to 0 lwarx r5,0,r3 #try again, reserving cmpwi r5,0 #(likely to succeed) bne loop #try to store nonzero stwcx. r4,0,r3 # bne- loop #loop if lost reservation [...] E.2.5 Test and Set This version of the test and set primitive atomically loads a word from memory, ensures that the word in memory is a nonzero value, and sets CR0[EQ] according to whether the value loaded is zero. In this example, it is assumed that the address of the word to be tested is in GPR3, the new value (nonzero) is in GPR4, and the old value is returned in GPR5. loop: lwarx r5,0,r3 #load and reserve cmpwi r5, 0 #done if word bne $+12 #not equal to 0 stwcx. r4,0,r3 #try to store non-zero bne- loop #loop if lost reservation </quote> regards, alexander.