[ofa-general] [PATCH 5/7 V3] osm: QoS - adding QoS policy options
Hal Rosenstock
hal.rosenstock at gmail.com
Thu Aug 30 04:04:26 PDT 2007
On 8/29/07, Yevgeny Kliteynik <kliteyn at dev.mellanox.co.il> wrote:
> Hal Rosenstock wrote:
> > On 8/29/07, Yevgeny Kliteynik <kliteyn at dev.mellanox.co.il> wrote:
> >> Hal Rosenstock wrote:
> >>> On 8/29/07, Yevgeny Kliteynik <kliteyn at dev.mellanox.co.il> wrote:
> >>>> Hal Rosenstock wrote:
> >>>>> On 8/27/07, Yevgeny Kliteynik <kliteyn at dev.mellanox.co.il> wrote:
> >>>>>> Adding QoS policy file option to OpenSM optionsm and
> >>>>>> QoS Policy field to subn object.
> >>>>>>
> >>>>>> Signed-off-by: Yevgeny Kliteynik <kliteyn at dev.mellanox.co.il>
> >>>>>> ---
> >>>>>> opensm/include/opensm/osm_base.h | 17 +++++++++++++++++
> >>>>>> opensm/include/opensm/osm_subnet.h | 13 +++++++++++++
> >>>>>> opensm/opensm/main.c | 12 +++++++++++-
> >>>>>> opensm/opensm/osm_subnet.c | 10 +++++++++-
> >>>>>> 4 files changed, 50 insertions(+), 2 deletions(-)
> >>>>>>
> >>>>>> diff --git a/opensm/include/opensm/osm_base.h b/opensm/include/opensm/osm_base.h
> >>>>>> index 545779b..0a35e22 100644
> >>>>>> --- a/opensm/include/opensm/osm_base.h
> >>>>>> +++ b/opensm/include/opensm/osm_base.h
> >>>>>> @@ -224,6 +224,23 @@ BEGIN_C_DECLS
> >>>>>> #define OSM_DEFAULT_PARTITION_CONFIG_FILE "/etc/ofa/opensm-partitions.conf"
> >>>>>> #endif
> >>>>>> /***********/
> >>>>>> +
> >>>>>> +/****d* OpenSM: Base/OSM_DEFAULT_QOS_POLICY_FILE
> >>>>>> +* NAME
> >>>>>> +* OSM_DEFAULT_QOS_POLICY_FILE
> >>>>>> +*
> >>>>>> +* DESCRIPTION
> >>>>>> +* Specifies the default QoS policy file name
> >>>>>> +*
> >>>>>> +* SYNOPSIS
> >>>>>> +*/
> >>>>>> +#ifdef __WIN__
> >>>>>> +#define OSM_DEFAULT_QOS_POLICY_FILE strcat(GetOsmCachePath(), "osm-qos-policy.conf")
> >>>>>> +#else
> >>>>>> +#define OSM_DEFAULT_QOS_POLICY_FILE "/etc/ofa/opensm-qos-policy.conf"
> >>>>>> +#endif
> >>>>>> +/***********/
> >>>>>> +
> >>>>>> /****d* OpenSM: Base/OSM_DEFAULT_SWEEP_INTERVAL_SECS
> >>>>>> * NAME
> >>>>>> * OSM_DEFAULT_SWEEP_INTERVAL_SECS
> >>>>>> diff --git a/opensm/include/opensm/osm_subnet.h b/opensm/include/opensm/osm_subnet.h
> >>>>>> index a7543dc..5e8b522 100644
> >>>>>> --- a/opensm/include/opensm/osm_subnet.h
> >>>>>> +++ b/opensm/include/opensm/osm_subnet.h
> >>>>>> @@ -68,6 +68,7 @@ BEGIN_C_DECLS
> >>>>>> #define OSM_SUBNET_VECTOR_GROW_SIZE 1
> >>>>>> #define OSM_SUBNET_VECTOR_CAPACITY 256
> >>>>>> struct _osm_opensm_t;
> >>>>>> +struct _osm_qos_policy_t;
> >>>>>>
> >>>>>> /****h* OpenSM/Subnet
> >>>>>> * NAME
> >>>>>> @@ -257,6 +258,7 @@ typedef struct _osm_subn_opt {
> >>>>>> char *partition_config_file;
> >>>>>> boolean_t no_partition_enforcement;
> >>>>>> boolean_t no_qos;
> >>>>>> + char *qos_policy_file;
> >>>>>> boolean_t accum_log_file;
> >>>>>> char *console;
> >>>>>> uint16_t console_port;
> >>>>>> @@ -398,6 +400,13 @@ typedef struct _osm_subn_opt {
> >>>>>> * specified the log file will be truncated upon reaching
> >>>>>> * this limit.
> >>>>>> *
> >>>>>> +* no_qos
> >>>>>> +* Boolean that specifies whether the OpenSM QoS functionality
> >>>>>> +* should be off or on.
> >>>>>> +*
> >>>>>> +* qos_policy_file
> >>>>>> +* Name of the QoS policy file.
> >>>>>> +*
> >>>>> Are these options mutually exclusive or not ?
> >>>> If QoS in OpenSM is off, the qos_policy_file will be ignored
> >>> But aren't there two QoSs ? Can't the old QoS be run without this ? I
> >>> think that is a requirement.
> >> OK, there are three ways we want this thing to work:
> >> 1. QoS is off
> >> 2. The old QoS is on but w/o policy file
> >> 3. The old QoS is on, plus reading policy file
> >>
> >> The first option is clear: if a user doesn't turns QoS on (-Q), QoS is off as before.
> >>
> >> Second and third options: if QoS is on, OpenSM looks for policy file in the default
> >> location or in other location that was provided by user. If the file is not found,
> >> QoS works as before.
> >
> > This sounds OK to me and is my first preference.
> >
> >> Do we want to add additional option for "enhanced" QoS?
> >> If so, we will have three QoS-ralated command line options:
> >> - option for turning the QoS on (currently -Q)
> >> - option to turn the new QoS on (some new letter - must get
> >> one quick before they all run out... :)
> >> - option for policy file location if differs from default (currently -Y)
> >
> > This seems like the least preferable to me. Also, would need to deal
> > with both on which seems to mean use new QoS.
> >
> >> Alternatively, we can turn -Q option into levels:
> >> -Q 0: QoS is off (default)
> >> -Q 1: old QoS is on
> >> -Q 2: old QoS plus reading policy file
> >
> > This one also seems OK to me (second preference).
> >
> > Anyone else with an opinion on this ? Sasha ?
> >
> > Also, what about --enable-qos-policy build option for those who don't
> > want to include this in their build ?
>
> Do you mean permanently, or just until the new QoS will be stable?
> Note that if someone gets ofed rpm, he gets the generated files in this
> rpm, so he doesn't need les or yacc, so this option would be useful only
> for machines that have git but don't have lex or yacc.
Didn't realize this.
How are the use of flex and bison dealt with in Windows ?
> Do you think it's a real issue?
Aside from the ongoing issues, there's always the OpenSM size issue as
Sasha mentioned which is an issue for some. This issue is larger than
just this feature.
-- Hal
> -- Yevgeny
>
> > -- Hal
> >
> >> -- Yevgeny
> >>
> >>>>>> * accum_log_file
> >>>>>> * If TRUE (default) - the log file will be accumulated.
> >>>>>> * If FALSE - the log file will be erased before starting current opensm run.
> >>>>>> @@ -551,6 +560,7 @@ typedef struct _osm_subn {
> >>>>>> ib_net64_t sm_port_guid;
> >>>>>> uint8_t sm_state;
> >>>>>> osm_subn_opt_t opt;
> >>>>>> + struct _osm_qos_policy_t *p_qos_policy;
> >>>>>> uint16_t max_unicast_lid_ho;
> >>>>>> uint16_t max_multicast_lid_ho;
> >>>>>> uint8_t min_ca_mtu;
> >>>>>> @@ -619,6 +629,9 @@ typedef struct _osm_subn {
> >>>>>> * opt
> >>>>>> * Subnet options structure contains site specific configuration.
> >>>>>> *
> >>>>>> +* p_qos_policy
> >>>>>> +* Subnet QoS policy structure.
> >>>>>> +*
> >>>>>> * max_unicast_lid_ho
> >>>>>> * The minimal max unicast lid reported by all switches
> >>>>>> *
> >>>>>> diff --git a/opensm/opensm/main.c b/opensm/opensm/main.c
> >>>>>> index e2541d8..599c4dc 100644
> >>>>>> --- a/opensm/opensm/main.c
> >>>>>> +++ b/opensm/opensm/main.c
> >>>>>> @@ -269,6 +269,10 @@ void show_usage(void)
> >>>>>> " The default name is \'"
> >>>>>> OSM_DEFAULT_PARTITION_CONFIG_FILE "\'.\n\n");
> >>>>>> printf("-Q\n" "--qos\n" " This option enables QoS setup.\n\n");
> >>>>>> + printf("-Y\n"
> >>>>>> + "--qos_policy_file\n"
> >>>>>> + " This option defines the optional QoS policy file.\n"
> >>>>>> + " The default name is \'" OSM_DEFAULT_QOS_POLICY_FILE "\'.\n\n");
> >>>>>> printf("-N\n"
> >>>>>> "--no_part_enforce\n"
> >>>>>> " This option disables partition enforcement on switch external ports.\n\n");
> >>>>>> @@ -580,7 +584,7 @@ int main(int argc, char *argv[])
> >>>>>> char *ignore_guids_file_name = NULL;
> >>>>>> uint32_t val;
> >>>>>> const char *const short_option =
> >>>>>> - "i:f:ed:g:l:L:s:t:a:u:R:zM:U:S:P:NBIQvVhorcyxp:n:q:k:C:";
> >>>>>> + "i:f:ed:g:l:L:s:t:a:u:R:zM:U:S:P:Y:NBIQvVhorcyxp:n:q:k:C:";
> >>>>>>
> >>>>>> /*
> >>>>>> In the array below, the 2nd parameter specifies the number
> >>>>>> @@ -604,6 +608,7 @@ int main(int argc, char *argv[])
> >>>>>> {"Pconfig", 1, NULL, 'P'},
> >>>>>> {"no_part_enforce", 0, NULL, 'N'},
> >>>>>> {"qos", 0, NULL, 'Q'},
> >>>>>> + {"qos_policy_file", 1, NULL, 'Y'},
> >>>>>> {"maxsmps", 1, NULL, 'n'},
> >>>>>> {"console", 1, NULL, 'q'},
> >>>>>> {"V", 0, NULL, 'V'},
> >>>>>> @@ -823,6 +828,11 @@ int main(int argc, char *argv[])
> >>>>>> opt.no_qos = FALSE;
> >>>>>> break;
> >>>>>>
> >>>>>> + case 'Y':
> >>>>>> + opt.qos_policy_file = optarg;
> >>>>>> + printf(" QoS policy file \'%s\'\n", optarg);
> >>>>>> + break;
> >>>>>> +
> >>>>> There should also be an update to the OpenSM man page for this.
> >>>> That and a text file with description of policy file syntax
> >>> Good. Thanks.
> >>>
> >>> -- Hal
> >>>
> >>>> -- Yevgeny
> >>>>
> >>>>> -- Hal
> >>>>>
> >>>>>> case 'y':
> >>>>>> opt.exit_on_fatal = FALSE;
> >>>>>> printf(" Staying on fatal initialization errors\n");
> >>>>>> diff --git a/opensm/opensm/osm_subnet.c b/opensm/opensm/osm_subnet.c
> >>>>>> index 818d73f..4162522 100644
> >>>>>> --- a/opensm/opensm/osm_subnet.c
> >>>>>> +++ b/opensm/opensm/osm_subnet.c
> >>>>>> @@ -452,6 +452,7 @@ void osm_subn_set_default_opt(IN osm_subn_opt_t * const p_opt)
> >>>>>> p_opt->partition_config_file = OSM_DEFAULT_PARTITION_CONFIG_FILE;
> >>>>>> p_opt->no_partition_enforcement = FALSE;
> >>>>>> p_opt->no_qos = TRUE;
> >>>>>> + p_opt->qos_policy_file = OSM_DEFAULT_QOS_POLICY_FILE;
> >>>>>> p_opt->accum_log_file = TRUE;
> >>>>>> p_opt->port_profile_switch_nodes = FALSE;
> >>>>>> p_opt->pfn_ui_pre_lid_assign = NULL;
> >>>>>> @@ -1178,6 +1179,9 @@ ib_api_status_t osm_subn_parse_conf_file(IN osm_subn_opt_t * const p_opts)
> >>>>>>
> >>>>>> opts_unpack_boolean("no_qos", p_key, p_val, &p_opts->no_qos);
> >>>>>>
> >>>>>> + opts_unpack_charp("qos_policy_file",
> >>>>>> + p_key, p_val, &p_opts->qos_policy_file);
> >>>>>> +
> >>>>>> opts_unpack_boolean("accum_log_file",
> >>>>>> p_key, p_val, &p_opts->accum_log_file);
> >>>>>>
> >>>>>> @@ -1541,7 +1545,11 @@ ib_api_status_t osm_subn_write_conf_file(IN osm_subn_opt_t * const p_opts)
> >>>>>> fprintf(opts_file,
> >>>>>> "#\n# QoS OPTIONS\n#\n"
> >>>>>> "# Disable QoS setup\n"
> >>>>>> - "no_qos %s\n\n", p_opts->no_qos ? "TRUE" : "FALSE");
> >>>>>> + "no_qos %s\n\n"
> >>>>>> + "# QoS policy file to be used\n"
> >>>>>> + "qos_policy_file %s\n\n",
> >>>>>> + p_opts->no_qos ? "TRUE" : "FALSE",
> >>>>>> + p_opts->qos_policy_file);
> >>>>>>
> >>>>>> subn_dump_qos_options(opts_file,
> >>>>>> "QoS default options", "qos",
> >>>>>> --
> >>>>>> 1.5.1.4
> >>>>>>
> >>>>>> _______________________________________________
> >>>>>> general mailing list
> >>>>>> general at lists.openfabrics.org
> >>>>>> http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
> >>>>>>
> >>>>>> To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
> >>>>>>
> >>
> >
>
>
More information about the general
mailing list