[openib-general] Re: kdapltest server on x86_64

[Hal Rosenstock]

On Tue, 2005-06-14 at 18:42, James Lentini wrote:
> I assume that kdapltest still reports that dat_psp_create fails due to 
> an invalid IA handle.

Yes. (I thought the added debug messages didn't change any behavior).
BTW, Tom found it and his fix works.

-- Hal

> Based on the trace:
> 
> On Tue, 14 Jun 2005, Hal Rosenstock wrote:
> 
> > On Tue, 2005-06-14 at 17:38, James Lentini wrote:
> >> Can you retest with revision 2602? I've added some debug prints that
> >> should tell us if the IA handle being passed to dapl_psp_create is
> >> really invalid.
> >
> > DAT: dat_ia_openv(name:mthca0a, async_qlen:8, ver:1.2, thrd_safe:0)
> > kDAPL: dapl_ia_open (mthca0a, 8, ffff81003f3b99c0, ffff81003f3b9990)
> > kDAPL: dapl_ia_open () returns 0x0
> > kDAPL: dapl_pz_create(ffff81003cac0080, ffff81003f3b9998)
> > kDAPL: dapl_evd_kcreate(ffff81003cac0080, 8, 1, ffff810012ee9e08, 20, ffff81003f3b99a0)
> > kDAPL: dapl_evd_kcreate(ffff81003cac0080, 8, 1, ffff810012ee9e08, A0, ffff81003f3b99a8)
> > kDAPL: dapl_evd_kcreate(ffff81003cac0080, 8, 1, ffff810012ee9e08, 10, ffff81003f3b99b8)
> > kDAPL: dapl_evd_kcreate(ffff81003cac0080, 8, 1, ffff810012ee9e08, 40, ffff81003f3b99b0)
> > kDAPL: dapl_ep_create (ffff81003cac0080, ffff81003f3b9800, ffff81001f6a6080, ffff81001776bac0, ffff81001776b5c0, 0000000000000000, ffff81003f3b99d0)
> > kDAPL: created qp_hndl = ffff81001f6e1800, qpn = 20406
> > kDAPL: dapl_psp_create(ffff81003cac0080, B0DE, ffff81001776b9c0, 0, ffff81003f3b99d8)
> 
> The correct IA is being passed to dapl_psp_create(). Since 
> dapl_psp_create's IA is clearly not null, the test on line 512 of 
> dapl_sp.c will not succeed.
> 
> There is no other place in the execution of dapl_psp_create where an 
> invalid IA handle error code is returned. 
> dapl_ib_setup_conn_listener() did return error values without using 
> the DAT_ERROR() macro to fill in the subtype. I wouldn't expect this 
> to cause any problems but I've updated the code to use DAT_ERROR() for 
> consistency.
> 
> Just to make sure that dat_str_error() is correctly parsing the error 
> code, I updated the dat_str_error() debug print.
> 
> Could you please retesting with revision 2605?
> 
> > DAT: dat_strerror(major:ffff810012ee9e38, minor:ffff810012ee9e40) called
> > kDAPL: dapl_ep_disconnect(ffff810028a49c00, 0)
> > kDAPL: dapl_ep_disconnect () returns 0x8007002e
> > DAT: dat_strerror(major:ffff810012ee9e38, minor:ffff810012ee9e40) called
> > kDAPL: dapl_ep_free (ffff810028a49c00)
> > kDAPL: dapl_ep_disconnect(ffff810028a49c00, 0)
> > kDAPL: dapl_ep_disconnect () returns 0x8007002e
> > kDAPL: dapl_ep_free: state 0, ep ffff810028a49c00 qp_state 1 qp 1f6e1800
> > kDAPL: dapl_evd_free (ffff81001776b5c0)
> > kDAPL: dapl_evd_free () returns 0x0
> > kDAPL: dapl_evd_free (ffff81001776b9c0)
> > kDAPL: dapl_evd_free () returns 0x0
> > kDAPL: dapl_evd_free (ffff81001776bac0)
> > kDAPL: dapl_evd_free () returns 0x0
> > kDAPL: dapl_evd_free (ffff81001f6a6080)
> > kDAPL: dapl_evd_free () returns 0x0
> > kDAPL: dapl_pz_free(ffff81003f3b9800)
> > DAT: dat_ia_close(ia:3cac0080, flags:1)
> > kDAPL: dapl_ia_query (ffff81003cac0080, 0000000000000000, 0000000000000000, ffff810012ee9c28)
> > kDAPL: dapl_ia_query () returns 0x0
> > kDAPL: dapl_ia_close (ffff81003cac0080, 1)
> > kDAPL: dapl_evd_free (ffff81001f6a6380)
> > kDAPL: dapl_evd_free () returns 0x0
> >
> >