Hi Neeraj, Thank you for your explanation! The lwn article is very helpful to me! Thanks, Alan > 2023年4月17日 下午1:28，Neeraj Upadhyay <quic_neeraju@xxxxxxxxxxx> 写道： > > Hi Alan, > > The reason for a different implementation looks to be due to the fact that the C standard does not specify the result of an operation that > overflows a signed integer. This is described in lwn article: https://lwn.net/Articles/511259/ . On 32-bit systems, overflow would be common for counters used by RCU. So, RCU subsystem uses unsigned integers checks - ULONG_CMP_GE , ULONG_CMP_LT , to avoid any compiler optimization, which would cause results of these checks to have a different result, in the event of counter overflows. > > > Thanks > Neeraj > > > On 4/16/2023 2:19 PM, Alan Huang wrote: >> Hi, >> I noticed that the definition of ULONG_CMP_LE in rcupdate.h is different with time_before. >> I can’t figure it out that why their definition is different. Does the two definition are the same thing? >> Thanks, >> Alan
