linux: Allow actually using CONFIG_XEN_BLKDEV_TAP=m
PATCH 1/1 linux: Allow actually using CONFIG_XEN_BLKDEV_TAP=m
Again, this got broken. It would be really nice if regression testing included building as a module everything that can be configured as such...
As usual, written and tested against 2.6.23 and made apply against 2.6.18 without further testing.
arch/i386/mm/hypervisor.c +1 -0
@@ -56,6 +56,7 @@ void xen_l1_entry_update(pte_t *ptr, pte u.val =__pte_val (val);BUG_ON (HYPERVISOR_mmu_update (&u,1 ,NULL ,DOMID_SELF ) <0 ); } +EXPORT_SYMBOL_GPL (xen_l1_entry_update);void xen_l2_entry_update (pmd_t *ptr,pmd_t val) {
drivers/xen/blktap/blktap.c +1 -7
@@ -116,13 +116,7 @@ typedef struct tap_blkif {static struct tap_blkif *tapfds[MAX_TAP_DEV ];static int blktap_next_minor; -static int __init set_blkif_reqs (char *str) -{ -get_option (&str, &blkif_reqs); -return 1 ; -} -__setup ("blkif_reqs=" , set_blkif_reqs); - +module_param (blkif_reqs, int,0 );/* Run-time switchable: /sys/module/blktap/parameters/ */ static unsigned int log_stats =0 ;static unsigned int debug_lvl =0 ;
Quoted 3 lines · Jan Beulich
Again, this got broken. It would be really nice if regression testing included building as a module everything that can be configured as such...
I recently discovered that the Linux build process has some
new options these days. Most useful and relevant here is:
make allmodconfig
make bzImage
The first one attempts to build *everything* as a module. There
are other targets as described in the top-level README which are new - or at least new to me..
Steve