lkml.org 
[lkml]   [2018]   [Mar]   [21]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
Subject[PATCH] target: Use WARNON_NON_RT(!irqs_disabled())
Hi,

We got a report where this WARN_ON got triggered:

[ 7552.799997] ------------[ cut here ]------------
[ 7552.800016] WARNING: CPU: 7 PID: 1090 at drivers/target/target_core_transport.c:3009 __transport_check_aborted_status+0x153/0x190 [target_core_mod]
[ 7552.800037] Modules linked in: target_core_user uio target_core_pscsi target_core_file target_core_iblock ib_srpt ib_srp scsi_transport_srp scsi_tgt xt_CHECKSUM iptable_mangle ipt_MASQUERADE nf_nat_masquerade_ipv4 iptable_nat nf_nat_ipv4 nf_nat nf_conntrack_ipv4 nf_defrag_ipv4 xt_conntrack nf_conntrack ipt_REJECT nf_reject_ipv4 tun ebtable_filter ebtables ip6table_filter ip6_tables iptable_filter bridge stp llc ib_isert iscsi_target_mod target_core_mod ib_ucm rpcrdma mlx5_ib sunrpc rdma_ucm ib_uverbs ib_iser rdma_cm iw_cm libiscsi ib_umad ib_ipoib scsi_transport_iscsi ib_cm sb_edac intel_powerclamp coretemp intel_rapl iosf_mbi crc32_pclmul ghash_clmulni_intel aesni_intel lrw gf128mul glue_helper ablk_helper cryptd iTCO_wdt iTCO_vendor_support hfi1 ipmi_ssif sg rdmavt ib_core hpilo hpwdt pcspkr ipmi_si
[ 7552.800055] ipmi_devintf ipmi_msghandler wmi acpi_power_meter ioatdma dca shpchp pcc_cpufreq lpc_ich ip_tables xfs libcrc32c mgag200 i2c_algo_bit drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops ttm sd_mod crc_t10dif crct10dif_generic crct10dif_pclmul crct10dif_common crc32c_intel serio_raw ata_generic pata_acpi mlx5_core ata_piix tg3 drm devlink libata i2c_core ptp hpsa scsi_transport_sas pps_core dm_mirror dm_region_hash dm_log dm_mod [last unloaded: ib_srpt]
[ 7552.800058] CPU: 7 PID: 1090 Comm: kworker/7:1H Not tainted 3.10.0-768.rt56.699.el7.x86_64 #1
[ 7552.800058] Hardware name: HP ProLiant DL380p Gen8, BIOS P70 11/14/2013
[ 7552.800066] Workqueue: ib-comp-wq ib_cq_poll_work [ib_core]
[ 7552.800067] Call Trace:
[ 7552.800075] [<ffffffffb76cd055>] dump_stack+0x19/0x1b
[ 7552.800078] [<ffffffffb70807bb>] __warn+0xfb/0x120
[ 7552.800080] [<ffffffffb70808fd>] warn_slowpath_null+0x1d/0x20
[ 7552.800085] [<ffffffffc0ab3983>] __transport_check_aborted_status+0x153/0x190 [target_core_mod]
[ 7552.800091] [<ffffffffc0ab5c04>] target_execute_cmd+0x34/0x2e0 [target_core_mod]
[ 7552.800096] [<ffffffffc0ab5fc2>] transport_generic_new_cmd+0x112/0x240 [target_core_mod]
[ 7552.800100] [<ffffffffc0ab6132>] transport_handle_cdb_direct+0x42/0x90 [target_core_mod]
[ 7552.800105] [<ffffffffc0ab62cd>] target_submit_cmd_map_sgls+0x14d/0x210 [target_core_mod]
[ 7552.800107] [<ffffffffc09c15b4>] srpt_handle_new_iu+0x254/0x660 [ib_srpt]
[ 7552.800109] [<ffffffffc09c1bc8>] srpt_recv_done+0x38/0x60 [ib_srpt]
[ 7552.800113] [<ffffffffc07a5fb5>] __ib_process_cq+0x65/0xe0 [ib_core]
[ 7552.800118] [<ffffffffc07a60a0>] ib_cq_poll_work+0x20/0x60 [ib_core]
[ 7552.800120] [<ffffffffb70a4336>] process_one_work+0x176/0x4a0
[ 7552.800121] [<ffffffffb70a50ec>] worker_thread+0x16c/0x3f0
[ 7552.800123] [<ffffffffb70a4f80>] ? manage_workers.isra.36+0x2b0/0x2b0
[ 7552.800125] [<ffffffffb70ac62f>] kthread+0xcf/0xe0
[ 7552.800139] [<ffffffffb70ac560>] ? kthread_worker_fn+0x170/0x170
[ 7552.800151] [<ffffffffb76dd1d8>] ret_from_fork+0x58/0x90
[ 7552.800153] [<ffffffffb70ac560>] ? kthread_worker_fn+0x170/0x170
[ 7552.800154] ---[ end trace 0000000000000002 ]---
[ 7554.164964] srpt/0xf4521403000e4aa0f4521403000e4ad0: Unsupported SCSI Opcode 0xa3, sending CHECK_CONDITION.
[ 7554.231254] srpt/0xf4521403000e4aa0f4521403000e4ad0: Unsupported SCSI Opcode 0xa3, sending CHECK_CONDITION.
[ 7554.294860] srpt/0xf4521403000e4aa0f4521403000e4ad0: Unsupported SCSI Opcode 0xa3, sending CHECK_CONDITION.
[ 7554.360810] srpt/0xf4521403000e4aa0f4521403000e4ad0: Unsupported SCSI Opcode 0xa3, sending CHECK_CONDITION.
[ 7554.421867] srpt/0xf4521403000e4aa0f4521403000e4ad0: Unsupported SCSI Opcode 0xa3, sending CHECK_CONDITION.
[ 7554.485931] srpt/0xf4521403000e4aa0f4521403000e4ad0: Unsupported SCSI Opcode 0xa3, sending CHECK_CONDITION.
[ 7554.546909] srpt/0xf4521403000e4aa0f4521403000e4ad0: Unsupported SCSI Opcode 0xa3, sending CHECK_CONDITION.
[ 7554.607820] srpt/0xf4521403000e4aa0f4521403000e4ad0: Unsupported SCSI Opcode 0xa3, sending CHECK_CONDITION.
[ 7554.671883] srpt/0xf4521403000e4aa0f4521403000e4ad0: Unsupported SCSI Opcode 0xa3, sending CHECK_CONDITION.
[ 7554.730826] srpt/0xf4521403000e4aa0f4521403000e4ad0: Unsupported SCSI Opcode 0xa3, sending CHECK_CONDITION.

static int __transport_check_aborted_status(struct se_cmd *cmd, int send_status)
__releases(&cmd->t_state_lock)
__acquires(&cmd->t_state_lock)
{
int ret;

assert_spin_locked(&cmd->t_state_lock);
WARN_ON_ONCE_NONRT(!irqs_disabled());
<SNIP>

And it is called just from these two places:

int transport_check_aborted_status(struct se_cmd *cmd, int send_status)
{
int ret;

spin_lock_irq(&cmd->t_state_lock);
ret = __transport_check_aborted_status(cmd, send_status);
spin_unlock_irq(&cmd->t_state_lock);

return ret;
}
EXPORT_SYMBOL(transport_check_aborted_status);

And:

void target_execute_cmd(struct se_cmd *cmd)
{
/*
* Determine if frontend context caller is requesting the stopping of
* this command for frontend exceptions.
*
* If the received CDB has aleady been aborted stop processing it here.
*/
spin_lock_irq(&cmd->t_state_lock);
if (__transport_check_aborted_status(cmd, 1)) {
spin_unlock_irq(&cmd->t_state_lock);
return;
}
<SNIP>

Since cmd->t_state_lock becomes a sleeping spin lock on RT, that thing
triggers, turn it into a NONRT WARN_ON, a kernel built with this patch
passes the test case that lead to a BZ being filled for the kernel-rt
package:

https://bugzilla.redhat.com/show_bug.cgi?id=1512875

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

---

diff --git a/drivers/target/target_core_transport.c b/drivers/target/target_core_transport.c
index 4558f2e1fe1b..318453e7adfd 100644
--- a/drivers/target/target_core_transport.c
+++ b/drivers/target/target_core_transport.c
@@ -3204,7 +3204,7 @@ static int __transport_check_aborted_status(struct se_cmd *cmd, int send_status)
int ret;

assert_spin_locked(&cmd->t_state_lock);
- WARN_ON_ONCE(!irqs_disabled());
+ WARN_ON_ONCE_NONRT(!irqs_disabled());

if (!(cmd->transport_state & CMD_T_ABORTED))
return 0;
\
 
 \ /
  Last update: 2018-03-21 16:40    [W:0.181 / U:27.824 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site