From mboxrd@z Thu Jan 1 00:00:00 1970 From: Sasha Khapyorsky Subject: Re: [PATCH] opensm/osm_mcast_tbl.c: Fix multicast group removal Date: Fri, 30 Oct 2009 22:36:06 +0200 Message-ID: <20091030203606.GD5829@me> References: <20091030144919.GA26847@comcast.net> <20091030102649.fa06e288.weiny2@llnl.gov> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <20091030102649.fa06e288.weiny2-i2BcT+NCU+M@public.gmane.org> Sender: linux-rdma-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: Ira Weiny , Hal Rosenstock Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-Id: linux-rdma@vger.kernel.org On 10:26 Fri 30 Oct , Ira Weiny wrote: > This patch causes this assert again... > > OpenSM $ opensm: osm_mcast_tbl.c:104: osm_mcast_tbl_set: Assertion `mlid_ho <= p_tbl->max_mlid_ho' failed. > Aborted (core dumped) > > I thought the -1 was required... The problem is different: when osm_mcast_tbl_realloc() is called twice let's say with 'mlid_offset = 0' and later with 'mlid_offset = 1' tbl->max_mlid_ho still have MLID_START value because its update is skipped in this patch. Seems wrong to me. Sasha > > diff --git a/opensm/opensm/osm_mcast_tbl.c b/opensm/opensm/osm_mcast_tbl.c > index da106c6..67f083e 100644 > --- a/opensm/opensm/osm_mcast_tbl.c > +++ b/opensm/opensm/osm_mcast_tbl.c > @@ -144,7 +144,7 @@ int osm_mcast_tbl_realloc(IN osm_mcast_tbl_t * p_tbl, IN uintn_t mlid_offset) > size - p_tbl->mft_depth * (IB_MCAST_POSITION_MAX + 1) * IB_MCAST_MASK_SIZE / 8); > p_tbl->p_mask_tbl = p_mask_tbl; > p_tbl->mft_depth = mft_depth; > - p_tbl->max_mlid_ho = mlid_offset + IB_LID_MCAST_START_HO; > + p_tbl->max_mlid_ho = mlid_offset + IB_LID_MCAST_START_HO - 1; > return 0; > } > > But this did not fix it... I would dig further but I am kind of busy today, sorry. > > Ira > > > On Fri, 30 Oct 2009 09:49:19 -0500 > Hal Rosenstock wrote: > > > > > Revert de3171043b3cbe078e410351a00e5756509bd514 and > > some of 537d67b19a4b72d72341e529f611e4e30754f8fc > > > > In osm_mcast_tbl_realloc, don't change max_mlid_ho when mlid_offset < p_tbl->mft_depth > > Also, restore some checks when mlid_offset >= p_tbl->mft_depth > > These changes are needed for proper group removal > > > > Signed-off-by: Hal Rosenstock > > --- > > diff --git a/opensm/opensm/osm_mcast_tbl.c b/opensm/opensm/osm_mcast_tbl.c > > index 0a45904..da106c6 100644 > > --- a/opensm/opensm/osm_mcast_tbl.c > > +++ b/opensm/opensm/osm_mcast_tbl.c > > @@ -122,7 +122,7 @@ int osm_mcast_tbl_realloc(IN osm_mcast_tbl_t * p_tbl, IN uintn_t mlid_offset) > > uint16_t (*p_mask_tbl)[][IB_MCAST_POSITION_MAX]; > > > > if (mlid_offset < p_tbl->mft_depth) > > - goto done; > > + return 0; > > > > /* > > The number of bytes needed in the mask table is: > > @@ -144,7 +144,6 @@ int osm_mcast_tbl_realloc(IN osm_mcast_tbl_t * p_tbl, IN uintn_t mlid_offset) > > size - p_tbl->mft_depth * (IB_MCAST_POSITION_MAX + 1) * IB_MCAST_MASK_SIZE / 8); > > p_tbl->p_mask_tbl = p_mask_tbl; > > p_tbl->mft_depth = mft_depth; > > -done: > > p_tbl->max_mlid_ho = mlid_offset + IB_LID_MCAST_START_HO; > > return 0; > > } > > @@ -167,6 +166,9 @@ boolean_t osm_mcast_tbl_is_port(IN const osm_mcast_tbl_t * p_tbl, > > CL_ASSERT(mlid_ho <= p_tbl->max_mlid_ho); > > > > mlid_offset = mlid_ho - IB_LID_MCAST_START_HO; > > + if (mlid_offset >= p_tbl->mft_depth) > > + return FALSE; > > + > > mask_offset = port_num / IB_MCAST_MASK_SIZE; > > bit_mask = cl_ntoh16((uint16_t) > > (1 << (port_num % IB_MCAST_MASK_SIZE))); > > @@ -194,6 +196,8 @@ boolean_t osm_mcast_tbl_is_any_port(IN const osm_mcast_tbl_t * p_tbl, > > CL_ASSERT(mlid_ho <= p_tbl->max_mlid_ho); > > > > mlid_offset = mlid_ho - IB_LID_MCAST_START_HO; > > + if (mlid_offset >= p_tbl->mft_depth) > > + return FALSE; > > > > for (position = 0; position <= p_tbl->max_position; position++) > > result |= (*p_tbl->p_mask_tbl)[mlid_offset][position]; > > @@ -247,6 +251,8 @@ void osm_mcast_tbl_clear_mlid(IN osm_mcast_tbl_t * p_tbl, IN uint16_t mlid_ho) > > > > if (p_tbl->p_mask_tbl && mlid_ho <= p_tbl->max_mlid_ho) { > > mlid_offset = mlid_ho - IB_LID_MCAST_START_HO; > > + if (mlid_offset >= p_tbl->mft_depth) > > + return; > > for (i = 0; i <= p_tbl->max_position; i++) > > (*p_tbl->p_mask_tbl)[mlid_offset][i] = 0; > > } > > -- > > To unsubscribe from this list: send the line "unsubscribe linux-rdma" in > > the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org > > More majordomo info at http://*vger.kernel.org/majordomo-info.html > > > > > -- > Ira Weiny > Math Programmer/Computer Scientist > Lawrence Livermore National Lab > 925-423-8008 > weiny2-i2BcT+NCU+M@public.gmane.org > -- > To unsubscribe from this list: send the line "unsubscribe linux-rdma" in > the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org > More majordomo info at http://vger.kernel.org/majordomo-info.html > -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html