Re: 2.4.17 alignment issue sparc/ia64

2015-10-16 Thread Rainer Jung
Am 16.10.2015 um 08:25 schrieb Jacob Champion: On 10/15/2015 11:18 PM, Jacob Champion wrote: it looks like ap_init_scoreboard() doesn't try to maintain any particular alignment when it's assigning pointers from more_storage. Though one would think your compiler would be padding out the struct

Re: 2.4.17 alignment issue sparc/ia64

2015-10-16 Thread Jacob Champion
On 10/15/2015 08:53 PM, Eric Covener wrote: We recently merged 2.4.17 and saw some bus errors on hp/ia64 and solaris/sparc64. Selectively backing things out, it appears that the SO_REUSEPORT patch causes the worker_score to no longer (necessarily) be double-word aligned. I don't have any

Re: 2.4.17 alignment issue sparc/ia64

2015-10-16 Thread Yann Ylavic
On Fri, Oct 16, 2015 at 5:53 AM, Eric Covener wrote: > We recently merged 2.4.17 and saw some bus errors on hp/ia64 and > solaris/sparc64. Selectively backing things out, it appears that the > SO_REUSEPORT patch causes the worker_score to no longer (necessarily) > be

Re: 2.4.17 alignment issue sparc/ia64

2015-10-16 Thread Jacob Champion
On 10/15/2015 11:18 PM, Jacob Champion wrote: it looks like ap_init_scoreboard() doesn't try to maintain any particular alignment when it's assigning pointers from more_storage. Though one would think your compiler would be padding out the struct to a double-word multiple anyway. Hrm.

Re: 2.4.17 alignment issue sparc/ia64

2015-10-16 Thread Jim Jagielski
I seem to recall similar issues w/ the shm slotmem impl... > On Oct 16, 2015, at 8:35 AM, Rainer Jung wrote: > > Am 16.10.2015 um 13:54 schrieb Yann Ylavic: >> On Fri, Oct 16, 2015 at 10:02 AM, Yann Ylavic wrote: >>> >>> We should do something

Re: 2.4.17 alignment issue sparc/ia64

2015-10-16 Thread Yann Ylavic
On Fri, Oct 16, 2015 at 2:48 PM, Yann Ylavic wrote: > On Fri, Oct 16, 2015 at 2:35 PM, Rainer Jung wrote: >> >> I didn't yet have the time to reproduce and test your patch, but the >> APR_ALIGN((size),sizeof(void *)) align approach would not work.

Re: 2.4.17 alignment issue sparc/ia64

2015-10-16 Thread Yann Ylavic
On Fri, Oct 16, 2015 at 10:02 AM, Yann Ylavic wrote: > > We should do something like the following patch: > > Index: server/scoreboard.c > === > --- server/scoreboard.c(revision 1708095) > +++

Re: 2.4.17 alignment issue sparc/ia64

2015-10-16 Thread Rainer Jung
Am 16.10.2015 um 13:54 schrieb Yann Ylavic: On Fri, Oct 16, 2015 at 10:02 AM, Yann Ylavic wrote: We should do something like the following patch: Index: server/scoreboard.c === --- server/scoreboard.c

Re: 2.4.17 alignment issue sparc/ia64

2015-10-16 Thread Yann Ylavic
On Fri, Oct 16, 2015 at 2:35 PM, Rainer Jung wrote: > > I didn't yet have the time to reproduce and test your patch, but the > APR_ALIGN((size),sizeof(void *)) align approach would not work. The problem > here is that even or especially when building for 32 Bits and then

Re: 2.4.17 alignment issue sparc/ia64

2015-10-16 Thread Yann Ylavic
On Fri, Oct 16, 2015 at 3:08 PM, Rainer Jung wrote: > Am 16.10.2015 um 14:56 schrieb Yann Ylavic: >> >> On Fri, Oct 16, 2015 at 2:48 PM, Yann Ylavic wrote: >>> >>> On Fri, Oct 16, 2015 at 2:35 PM, Rainer Jung >>> wrote:

Re: 2.4.17 alignment issue sparc/ia64

2015-10-16 Thread Yann Ylavic
On Fri, Oct 16, 2015 at 3:16 PM, Yann Ylavic wrote: > On Fri, Oct 16, 2015 at 3:08 PM, Rainer Jung wrote: >> >> Wasn't the bus error occuring in >> >> ws->last_used = apr_time_now(); >> >> and the address is >> >> (dbx) print &(ws->last_used) >>

Re: 2.4.17 alignment issue sparc/ia64

2015-10-16 Thread Rainer Jung
Am 16.10.2015 um 14:56 schrieb Yann Ylavic: On Fri, Oct 16, 2015 at 2:48 PM, Yann Ylavic wrote: On Fri, Oct 16, 2015 at 2:35 PM, Rainer Jung wrote: I didn't yet have the time to reproduce and test your patch, but the

Re: 2.4.17 alignment issue sparc/ia64

2015-10-16 Thread Jim Jagielski
> > Yes but ws itself isn't aligned either: >(dbx) print ws >ws = 0x7bb00044 > which is IMHO the issue. > > Align ws and everything goes well (at least I think :p ). > It better! :)

Re: 2.4.17 alignment issue sparc/ia64

2015-10-16 Thread Eric Covener
On Fri, Oct 16, 2015 at 4:02 AM, Yann Ylavic wrote: > We should do something like the following patch: Promising so far with my two one-off testcases, putting it through a longer test now. Thanks!

Re: 2.4.17 alignment issue sparc/ia64

2015-10-16 Thread Yann Ylavic
On Fri, Oct 16, 2015 at 4:16 PM, Eric Covener wrote: > On Fri, Oct 16, 2015 at 4:02 AM, Yann Ylavic wrote: >> We should do something like the following patch: > > Promising so far with my two one-off testcases, putting it through a > longer test now.

2.4.17 alignment issue sparc/ia64

2015-10-15 Thread Eric Covener
We recently merged 2.4.17 and saw some bus errors on hp/ia64 and solaris/sparc64. Selectively backing things out, it appears that the SO_REUSEPORT patch causes the worker_score to no longer (necessarily) be double-word aligned. I'm able to hit the same thing with the httpd-2.4.x branch. I have a