linux-kernel - [PATCH] arcmsr: Fix possible sleep-in-atomic bugs in arcmsr_queue

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Message-Id: <1513085580-28705-1-git-send-email-baijiaju1990@gmail.com>
Date:   Tue, 12 Dec 2017 21:33:00 +0800
From:   Jia-Ju Bai <baijiaju1990@...il.com>
To:     support@...ca.com.tw, jejb@...ux.vnet.ibm.com,
        martin.petersen@...cle.com, keescook@...omium.org
Cc:     linux-scsi@...r.kernel.org, linux-kernel@...r.kernel.org,
        Jia-Ju Bai <baijiaju1990@....com>
Subject: [PATCH] arcmsr: Fix possible sleep-in-atomic bugs in arcmsr_queue_command

From: Jia-Ju Bai <baijiaju1990@....com>

The driver may sleep under a spinlock, and the function call paths are:
arcmsr_queue_command(acquire the spinlock)
  arcmsr_queue_command_lck
    arcmsr_handle_virtual_command
      arcmsr_iop_message_xfer
        arcmsr_iop_parking
          arcmsr_stop_adapter_bgrb
            arcmsr_hbaA_stop_bgrb
              arcmsr_hbaA_wait_msgint_ready
                msleep --> may sleep

arcmsr_queue_command(acquire the spinlock)
  arcmsr_queue_command_lck
    arcmsr_handle_virtual_command
      arcmsr_iop_message_xfer
        arcmsr_iop_parking
          arcmsr_stop_adapter_bgrb
            arcmsr_hbaB_stop_bgrb
              arcmsr_hbaB_wait_msgint_ready
                msleep --> may sleep

arcmsr_queue_command(acquire the spinlock)
  arcmsr_queue_command_lck
    arcmsr_handle_virtual_command
      arcmsr_iop_message_xfer
        arcmsr_iop_parking
          arcmsr_stop_adapter_bgrb
            arcmsr_hbaC_stop_bgrb
              arcmsr_hbaC_wait_msgint_ready
                msleep --> may sleep

arcmsr_queue_command(acquire the spinlock)
  arcmsr_queue_command_lck
    arcmsr_handle_virtual_command
      arcmsr_iop_message_xfer
        arcmsr_iop_parking
          arcmsr_stop_adapter_bgrb
            arcmsr_hbaD_stop_bgrb
              arcmsr_hbaD_wait_msgint_ready
                msleep --> may sleep

To fix them, msleep is replaced with mdelay.
These bugs are found by my static analysis tool and my code review.

Signed-off-by: Jia-Ju Bai <baijiaju1990@....com>
---
 drivers/scsi/arcmsr/arcmsr_hba.c |    8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/scsi/arcmsr/arcmsr_hba.c b/drivers/scsi/arcmsr/arcmsr_hba.c
index af032c4..31d94bd 100644
--- a/drivers/scsi/arcmsr/arcmsr_hba.c
+++ b/drivers/scsi/arcmsr/arcmsr_hba.c
@@ -347,7 +347,7 @@ static uint8_t arcmsr_hbaA_wait_msgint_ready(struct AdapterControlBlock *acb)
 				&reg->outbound_intstatus);
 			return true;
 		}
-		msleep(10);
+		mdelay(10);
 	} /* max 20 seconds */
 
 	return false;
@@ -367,7 +367,7 @@ static uint8_t arcmsr_hbaB_wait_msgint_ready(struct AdapterControlBlock *acb)
 					reg->drv2iop_doorbell);
 			return true;
 		}
-		msleep(10);
+		mdelay(10);
 	} /* max 20 seconds */
 
 	return false;
@@ -385,7 +385,7 @@ static uint8_t arcmsr_hbaC_wait_msgint_ready(struct AdapterControlBlock *pACB)
 				&phbcmu->outbound_doorbell_clear); /*clear interrupt*/
 			return true;
 		}
-		msleep(10);
+		mdelay(10);
 	} /* max 20 seconds */
 
 	return false;
@@ -403,7 +403,7 @@ static bool arcmsr_hbaD_wait_msgint_ready(struct AdapterControlBlock *pACB)
 				reg->outbound_doorbell);
 			return true;
 		}
-		msleep(10);
+		mdelay(10);
 	} /* max 20 seconds */
 	return false;
 }
-- 
1.7.9.5