Re: [PATCH 5/8] sg: add free list, rework locking

[Bart Van Assche <bvanassche@xxxxxxx>]

On Fri, 2018-10-19 at 15:55 -0400, Douglas Gilbert wrote:
+AD4 On 2018-10-19 11:22 a.m., Bart Van Assche wrote:
+AD4 +AD4 On Fri, 2018-10-19 at 02:24 -0400, Douglas Gilbert wrote:
+AD4 +AD4 +AD4 static void
+AD4 +AD4 +AD4 -sg+AF8-fill+AF8-request+AF8-table(struct sg+AF8-fd +ACo-sfp, struct sg+AF8-req+AF8-info +ACo-rinfo)
+AD4 +AD4 +AD4 +-sg+AF8-fill+AF8-request+AF8-table(struct sg+AF8-fd +ACo-sfp, struct sg+AF8-req+AF8-info +ACo-rinfo,
+AD4 +AD4 +AD4 +-                     int max+AF8-num)
+AD4 +AD4 +AD4   +AHs
+AD4 +AD4 +AD4          struct sg+AF8-request +ACo-srp+ADs
+AD4 +AD4 +AD4          int val+ADs
+AD4 +AD4 +AD4 -       unsigned int ms+ADs
+AD4 +AD4 +AD4   
+AD4 +AD4 +AD4          val +AD0 0+ADs
+AD4 +AD4 +AD4 -       list+AF8-for+AF8-each+AF8-entry(srp, +ACY-sfp-+AD4-rq+AF8-list, entry) +AHs
+AD4 +AD4 +AD4 -               if (val +AD4APQ SG+AF8-MAX+AF8-QUEUE)
+AD4 +AD4 +AD4 -                       break+ADs
+AD4 +AD4 +AD4 -               rinfo+AFs-val+AF0.req+AF8-state +AD0 srp-+AD4-done +- 1+ADs
+AD4 +AD4 +AD4 +-       list+AF8-for+AF8-each+AF8-entry(srp, +ACY-sfp-+AD4-rq+AF8-list, rq+AF8-entry) +AHs
+AD4 +AD4 +AD4 +-               if (val +AD4APQ max+AF8-num)
+AD4 +AD4 +AD4 +-                       return+ADs
+AD4 +AD4 
+AD4 +AD4 What protects the sfp-+AD4-rq+AF8-list against concurrent changes? It seems to me
+AD4 +AD4 like all other code that iterates over or modifies that list protects that
+AD4 +AD4 list with rq+AF8-list+AF8-lock?
+AD4 
+AD4 Bart,
+AD4 The function is called from sg+AF8-ioctl() case SG+AF8-GET+AF8-REQUEST+AF8-TABLE and at the
+AD4 call point the read+AF8-lock is held on rq+AF8-list+AF8-lock. Maybe I can add a comment
+AD4 above the function about the lock being held. +AFs-At least it is as the end
+AD4 of the patch series, and that is all I care about :-)+AF0

Hi Doug,

Thanks for the clarification. How about adding a +AF8AXw-must+AF8-hold() annotation?

Bart.