On Fri, 30 Mar 2018, Leon Romanovsky wrote: > On Wed, Mar 28, 2018 at 04:57:33PM -0500, Christopher Lameter wrote: > > On Tue, 27 Mar 2018, Leon Romanovsky wrote: > > > > > + > > > + mutex_lock(&stats->lock); > > > + msecs = jiffies_to_msecs(stats->lifespan); > > > + mutex_unlock(&stats->lock); > > > + > > > return sprintf(buf, "%d\n", msecs); > > > > Yuck. Can we use atomic64_t instead of u64 there? There are already ops > > defined for these types of scalars that ensure atomicity when using a > > variety of operations. The way this is going means that one needs to be > > aware of the special nature of the u64s here and manually take locks which > > is error prone. > > > > Plus you wont need a mutex with atomc64_t. > > > > Plus it falls back to sane code on 64 bit platforms where this is not > > needed. > Do you mean to use get_jiffies_64() and friends? No I mean that the stats should be atomic64_t if you want safe 32 bit support. You are here improvising safe handling of 64 bit entities in the driver subsystem of the kernel using a mutex. That is not the way to do things. You need to be using proper kernel infrastructure for handling 64 bit entities. -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
