Skip to contentSkip to conversation

ltp_tpci.c: fix a null pointer

PATCH In review Yang Xu 5 participants 9 messages Mar 18, 2020 – Apr 13, 2020

lore
Share & tools
lore

Yang Xu

author
#0 lore raw

PATCH 1/1 ltp_tpci.c: fix a null pointer

Since commit 3dd286bdbee ("ltp_tpci.c: Update legacy code"),
it introduced a warning as below:
../ltp_tpci.c: In function ?probe_pci_dev?:
../ltp_tpci.c:107:8: warning: ?dev? is used uninitialized in this function [-Wuninitialized]
struct pci_dev *dev;
dev = pci_get_domain_bus_and_slot(pci_domain_nr(dev->bus), bus, slot)

This will lead to system crash when we run this case because it triggers a null pointer. Use 0 domain intead of pci_domain_nr.

ps: I want to use a dymaic domain to fix it but failed. If someone know, please tell me. Also, this case failed when merging this patch because test 13 (test_assign_resources) report no space error as below:

[754930.757585] ltp_tpci: test-case 13
[754930.757585] ltp_tpci: assign resources
[754930.757585] ltp_tpci: assign resource #0
[754930.757586] ltp_tpci: name = 0000:00:08.0, flags = 262401, start 0xc140, end 0xc17f
[754930.757586] ltp_tpci: assign resource #1
[754930.757587] ltp_tpci: name = 0000:00:08.0, flags = 262656, start 0xfebd7000, end 0xfebd7fff
[754930.757587] ltp_tpci: assign resource #2
[754930.757587] ltp_tpci: name = 0000:00:08.0, flags = 0, start 0x0, end 0x0
[754930.757588] ltp_tpci: assign resource #3
[754930.757588] ltp_tpci: name = 0000:00:08.0, flags = 0, start 0x0, end 0x0
[754930.757588] ltp_tpci: assign resource #4
[754930.757589] ltp_tpci: name = 0000:00:08.0, flags = 538190348, start 0xfe80c000, end 0xfe80ffff
[754930.757593] virtio-pci 0000:00:08.0: BAR 4: no space for [mem size 0x00004000 64bit pref]
[754930.757594] virtio-pci 0000:00:08.0: BAR 4: failed to assign [mem size 0x00004000 64bit pref
Fixes 3dd286b ("ltp_tpci.c: Update legacy code")
---
 testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c +1 -1
index 7cbabfaa5..a57953db6 100644
@@ -104,7 +104,7 @@ static int probe_pci_dev(unsigned int bus, unsigned int slot)
 		ltp_pci.dev = NULL;
 	}
 
-	dev = pci_get_domain_bus_and_slot(pci_domain_nr(dev->bus), bus, slot);
+	dev = pci_get_domain_bus_and_slot(0, bus, slot);
 	if (!dev || !dev->driver)
 		return -ENODEV;
 
Signature · 1 line
2.18.1
author
#1 lore raw

PATCH 1/1 ltp_tpci.c: fix a null pointer

Hi

Does someone notice this(I guess this is a simple fix)?

Best Regards
Yang Xu

Quoted 48 lines

Since commit 3dd286bdbee ("ltp_tpci.c: Update legacy code"),
it introduced a warning as below:
../ltp_tpci.c: In function ?probe_pci_dev?:
../ltp_tpci.c:107:8: warning: ?dev? is used uninitialized in this function [-Wuninitialized]

   struct pci_dev *dev;
   dev = pci_get_domain_bus_and_slot(pci_domain_nr(dev->bus), bus, slot)

This will lead to system crash when we run this case because it triggers a null pointer. Use 0 domain intead of pci_domain_nr.

ps: I want to use a dymaic domain to fix it but failed. If someone know, please tell me. Also, this case failed when merging this patch because test 13 (test_assign_resources) report no space error as below:

[754930.757585] ltp_tpci: test-case 13
[754930.757585] ltp_tpci: assign resources
[754930.757585] ltp_tpci: assign resource #0
[754930.757586] ltp_tpci: name = 0000:00:08.0, flags = 262401, start 0xc140, end 0xc17f
[754930.757586] ltp_tpci: assign resource #1
[754930.757587] ltp_tpci: name = 0000:00:08.0, flags = 262656, start 0xfebd7000, end 0xfebd7fff
[754930.757587] ltp_tpci: assign resource #2
[754930.757587] ltp_tpci: name = 0000:00:08.0, flags = 0, start 0x0, end 0x0
[754930.757588] ltp_tpci: assign resource #3
[754930.757588] ltp_tpci: name = 0000:00:08.0, flags = 0, start 0x0, end 0x0
[754930.757588] ltp_tpci: assign resource #4
[754930.757589] ltp_tpci: name = 0000:00:08.0, flags = 538190348, start 0xfe80c000, end 0xfe80ffff
[754930.757593] virtio-pci 0000:00:08.0: BAR 4: no space for [mem size 0x00004000 64bit pref]
[754930.757594] virtio-pci 0000:00:08.0: BAR 4: failed to assign [mem size 0x00004000 64bit pref
Fixes 3dd286b ("ltp_tpci.c: Update legacy code")
---
  testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c +1 -1
index 7cbabfaa5..a57953db6 100644
@@ -104,7 +104,7 @@ static int probe_pci_dev(unsigned int bus, unsigned int slot)
  		ltp_pci.dev = NULL;
  	}
 
-	dev = pci_get_domain_bus_and_slot(pci_domain_nr(dev->bus), bus, slot);
+	dev = pci_get_domain_bus_and_slot(0, bus, slot);
  	if (!dev || !dev->driver)
  		return -ENODEV;
 
author
#4 lore raw

Hi Po-Hsu

Quoted 2 lines

We have encountered this null pointer dereference issue too, with this fix the test can be executed properly.

Thanks for your test.

Best Regards
Yang Xu

Quoted 4 lines
#5 lore raw

PATCH 1/1 ltp_tpci.c: fix a null pointer

In reply to Yang Xu

Greetings Yang,thanks for the patch.

Quoted 39 lines · Yang Xu

Since commit 3dd286bdbee ("ltp_tpci.c: Update legacy code"),
it introduced a warning as below:
../ltp_tpci.c: In function ?probe_pci_dev?:
../ltp_tpci.c:107:8: warning: ?dev? is used uninitialized in this function [-Wuninitialized]

   struct pci_dev *dev;
   dev = pci_get_domain_bus_and_slot(pci_domain_nr(dev->bus), bus, slot)

This will lead to system crash when we run this case because it triggers a null pointer. Use 0 domain intead of pci_domain_nr.

ps: I want to use a dymaic domain to fix it but failed. If someone know, please tell me. Also, this case failed when merging this patch because test 13 (test_assign_resources) report no space error as below:

[754930.757585] ltp_tpci: test-case 13
[754930.757585] ltp_tpci: assign resources
[754930.757585] ltp_tpci: assign resource #0
[754930.757586] ltp_tpci: name = 0000:00:08.0, flags = 262401, start 0xc140, end 0xc17f
[754930.757586] ltp_tpci: assign resource #1
[754930.757587] ltp_tpci: name = 0000:00:08.0, flags = 262656, start 0xfebd7000, end 0xfebd7fff
[754930.757587] ltp_tpci: assign resource #2
[754930.757587] ltp_tpci: name = 0000:00:08.0, flags = 0, start 0x0, end 0x0
[754930.757588] ltp_tpci: assign resource #3
[754930.757588] ltp_tpci: name = 0000:00:08.0, flags = 0, start 0x0, end 0x0
[754930.757588] ltp_tpci: assign resource #4
[754930.757589] ltp_tpci: name = 0000:00:08.0, flags = 538190348, start 0xfe80c000, end 0xfe80ffff
[754930.757593] virtio-pci 0000:00:08.0: BAR 4: no space for [mem size 0x00004000 64bit pref]
[754930.757594] virtio-pci 0000:00:08.0: BAR 4: failed to assign [mem size 0x00004000 64bit pref
Fixes 3dd286b ("ltp_tpci.c: Update legacy code")
---
  testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c +0 -0
index 7cbabfaa5..a57953db6 100644

testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c static int probe_pci_dev(unsigned int bus, unsigned int slot) · line 109

Quoted 7 lines
testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c +1 -1
@@ -104,7 +104,7 @@ static int probe_pci_dev(unsigned int bus, unsigned int slot)
  		ltp_pci.dev = NULL;
  	}
 
-	dev = pci_get_domain_bus_and_slot(pci_domain_nr(dev->bus), bus, slot);
+	dev = pci_get_domain_bus_and_slot(0, bus, slot);
  	if (!dev || !dev->driver)
  		return -ENODEV;

I was about to post the same fix on the mailing list - just tested in powerpc arch and it looks good to me.

It is safe to use 0 since function pci_get_bus_and_slot(unsigned int bus, unsigned int devfn) removed on kernel commit <5cf0c37a71da0f3> ("PCI: Remove pci_get_bus_and_slot() function") used to be a wrapper for pci_get_domain_bus_and_slot(0, bus, devfn).

Hence, this patch avoids the following Oops:

=====

[69614.978596] Unable to handle kernel paging request for data at 
address 0x00000010
[69614.978602] Faulting instruction address: 0xd000000003c200a4
[69614.978606] Oops: Kernel access of bad area, sig: 11 [#1]
[69614.978609] LE SMP NR_CPUS=2048 NUMA PowerNV
[69614.978613] Modules linked in: ltp_tpci(OE) vmac chacha20_generic 

poly1305_generic chacha20poly1305 snd_timer snd soundcore authenc pcrypt crypto_user sha3_generic salsa20_generic uinput can_raw can dummy veth n_gsm pps_ldisc ppp_synctty n_hdlc ppp_async ppp_generic slhc kvm_hv kvm_pr kvm binfmt_misc nfsv3 nfs_acl nfs lockd grace fscache tun brd vfat fat fuse overlay ext4 mbcache jbd2 loop sunrpc uio_pdrv_genirq ses xts ipmi_powernv enclosure ipmi_devintf uio scsi_transport_sas ibmpowernv ipmi_msghandler leds_powernv vmx_crypto powernv_rng powernv_op_panel ip_tables xfs libcrc32c sr_mod cdrom sd_mod sg ipr libata tg3 dm_mirror dm_region_hash dm_log dm_mod [last unloaded:
ltp_block_dev]

[69614.978653] Features: eBPF/sock
[69614.978657] CPU: 1 PID: 2292176 Comm: tpci Kdump: loaded Tainted: 
G??????? W? OE??? --------- -? - 4.18.0-187.el8.ppc64le #1
[69614.978662] NIP:? d000000003c200a4 LR: d000000003c2006c CTR: 
c000000000dc8c00
[69614.978666] REGS: c0000004be437a20 TRAP: 0300?? Tainted: G??????? W? 
OE??? --------- -? -? (4.18.0-187.el8.ppc64le)
[69614.978670] MSR:? 9000000000009033 <SF,HV,EE,ME,IR,DR,RI,LE>? CR: 
28002272? XER: 20000000
[69614.978676] CFAR: c000000000008934 DAR: 0000000000000010 DSISR: 

40000000 IRQMASK: 0
GPR00: d000000003c2006c c0000004be437ca0 d000000003c2a300 0000000000000000
GPR04: d000000003c214ba 0000000000000000 c0000007846448c9 0000000000000000
GPR08: c0000004be437d04 0000000000000000 0000000000000000 0000000000000000
GPR12: 0000000000002200 c0000007ffffee00 0000000000000000 0000000000000000
GPR16: 0000000000000000 000000001001cf98 000000001001cf78 000000001001cff0
GPR20: 000000001001d010 000000001001cf58 00007fff92643190 0000000010040530
GPR24: 000000001001d108 000000001001cf48 c0000004be437df0 0000000000000001
GPR28: fffffffffffffff2 0000000000000000 d000000003c22f80 0000000000000000

[69614.978711] NIP [d000000003c200a4] sys_bus_slot+0x8c/0x170 [ltp_tpci]
[69614.978715] LR [d000000003c2006c] sys_bus_slot+0x54/0x170 [ltp_tpci]
[69614.978718] Call Trace:
[69614.978721] [c0000004be437ca0] [d000000003c2006c] 
sys_bus_slot+0x54/0x170 [ltp_tpci] (unreliable)
[69614.978727] [c0000004be437d40] [c0000000008da5bc] 
dev_attr_store+0x3c/0x60
[69614.978732] [c0000004be437d60] [c0000000005fb908] 
sysfs_kf_write+0x68/0x80
[69614.978736] [c0000004be437d80] [c0000000005f9fb4] 
kernfs_fop_write+0x104/0x270
[69614.978741] [c0000004be437dd0] [c0000000004ff624] sys_write+0x134/0x3a0
[69614.978745] [c0000004be437e30] [c00000000000b388] system_call+0x5c/0x70
[69614.978748] Instruction dump:
[69614.978751] ebde8010 80a10064 e87e0000 78bdc622 54bf063e 2fa30000 
419e0014 48000e89
[69614.978757] e8410018 39200000 f93e0000 39200000 <e8690010> 48000dd1 
e8410018 7fe5fb78
[69614.978764] ---[ end trace 414b1270f579351f ]---

=====

Quoted 1 line
Signature · 4 lines
Desnes A. Nunes do Rosario
--------------------------------
Advisory Software Engineer - IBM
Virtual Onsite Engineer - Red Hat
author
#6 lore raw

PATCH 1/1 ltp_tpci.c: fix a null pointer

Hi Desnes

Quoted 65 lines

Greetings Yang,thanks for the patch.

Quoted 50 lines · Yang Xu

Since commit 3dd286bdbee ("ltp_tpci.c: Update legacy code"), it introduced a warning as below:
../ltp_tpci.c: In function ?probe_pci_dev?:
../ltp_tpci.c:107:8: warning: ?dev? is used uninitialized in this function [-Wuninitialized]
?? struct pci_dev *dev;
?? dev = pci_get_domain_bus_and_slot(pci_domain_nr(dev->bus), bus, slot)

This will lead to system crash when we run this case because it triggers a null pointer.
Use 0 domain intead of pci_domain_nr.

ps: I want to use a dymaic domain to fix it but failed. If someone know, please tell me. Also, this case failed when merging this patch because test 13 (test_assign_resources) report no space error as below:

[754930.757585] ltp_tpci: test-case 13
[754930.757585] ltp_tpci: assign resources
[754930.757585] ltp_tpci: assign resource #0
[754930.757586] ltp_tpci: name = 0000:00:08.0, flags = 262401, start 
0xc140, end 0xc17f
[754930.757586] ltp_tpci: assign resource #1
[754930.757587] ltp_tpci: name = 0000:00:08.0, flags = 262656, start 
0xfebd7000, end 0xfebd7fff
[754930.757587] ltp_tpci: assign resource #2
[754930.757587] ltp_tpci: name = 0000:00:08.0, flags = 0, start 0x0, 
end 0x0
[754930.757588] ltp_tpci: assign resource #3
[754930.757588] ltp_tpci: name = 0000:00:08.0, flags = 0, start 0x0, 
end 0x0
[754930.757588] ltp_tpci: assign resource #4
[754930.757589] ltp_tpci: name = 0000:00:08.0, flags = 538190348, 
start 0xfe80c000, end 0xfe80ffff
[754930.757593] virtio-pci 0000:00:08.0: BAR 4: no space for [mem size 
0x00004000 64bit pref]
[754930.757594] virtio-pci 0000:00:08.0: BAR 4: failed to assign [mem 

size 0x00004000 64bit pref

Fixes 3dd286b ("ltp_tpci.c: Update legacy code")

---
? testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c | 2 +- ? 1 file changed, 1 insertion(+), 1 deletion(-)

(quoted diff) +0 -0

a/testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c b/testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c index 7cbabfaa5..a57953db6 100644

testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c static int probe_pci_dev(unsigned int bus, · line 110

Quoted 8 lines
testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c +1 -1
@@ -104,7 +104,7 @@ static int probe_pci_dev(unsigned int bus, 
 unsigned int slot)
 ????????? ltp_pci.dev = NULL;
 ????? }
-??? dev = pci_get_domain_bus_and_slot(pci_domain_nr(dev->bus), bus, 
 slot);
+??? dev = pci_get_domain_bus_and_slot(0, bus, slot);
 ????? if (!dev || !dev->driver)
 ????????? return -ENODEV;

I was about to post the same fix on the mailing list - just tested in powerpc arch and it looks good to me.

Quoted 4 lines

It is safe to use 0 since function pci_get_bus_and_slot(unsigned int bus, unsigned int devfn) removed on kernel commit <5cf0c37a71da0f3> ("PCI: Remove pci_get_bus_and_slot() function") used to be a wrapper for pci_get_domain_bus_and_slot(0, bus, devfn).

Sorry for the late reply. Thanks for your review. I think this should be added into commit message, so make this fix more clear.

Best Regards
Yang Xu

Quoted 63 lines

Hence, this patch avoids the following Oops:

=====

[69614.978596] Unable to handle kernel paging request for data at 
address 0x00000010
[69614.978602] Faulting instruction address: 0xd000000003c200a4
[69614.978606] Oops: Kernel access of bad area, sig: 11 [#1]
[69614.978609] LE SMP NR_CPUS=2048 NUMA PowerNV
[69614.978613] Modules linked in: ltp_tpci(OE) vmac chacha20_generic 

poly1305_generic chacha20poly1305 snd_timer snd soundcore authenc pcrypt crypto_user sha3_generic salsa20_generic uinput can_raw can dummy veth n_gsm pps_ldisc ppp_synctty n_hdlc ppp_async ppp_generic slhc kvm_hv kvm_pr kvm binfmt_misc nfsv3 nfs_acl nfs lockd grace fscache tun brd vfat fat fuse overlay ext4 mbcache jbd2 loop sunrpc uio_pdrv_genirq ses xts ipmi_powernv enclosure ipmi_devintf uio scsi_transport_sas ibmpowernv ipmi_msghandler leds_powernv vmx_crypto powernv_rng powernv_op_panel ip_tables xfs libcrc32c sr_mod cdrom sd_mod sg ipr libata tg3 dm_mirror dm_region_hash dm_log dm_mod [last unloaded:
ltp_block_dev]

[69614.978653] Features: eBPF/sock
[69614.978657] CPU: 1 PID: 2292176 Comm: tpci Kdump: loaded Tainted: 
G??????? W? OE??? --------- -? - 4.18.0-187.el8.ppc64le #1
[69614.978662] NIP:? d000000003c200a4 LR: d000000003c2006c CTR: 
c000000000dc8c00
[69614.978666] REGS: c0000004be437a20 TRAP: 0300?? Tainted: G??????? W 
OE??? --------- -? -? (4.18.0-187.el8.ppc64le)
[69614.978670] MSR:? 9000000000009033 <SF,HV,EE,ME,IR,DR,RI,LE>? CR: 
28002272? XER: 20000000
[69614.978676] CFAR: c000000000008934 DAR: 0000000000000010 DSISR: 

40000000 IRQMASK: 0
GPR00: d000000003c2006c c0000004be437ca0 d000000003c2a300 0000000000000000
GPR04: d000000003c214ba 0000000000000000 c0000007846448c9 0000000000000000
GPR08: c0000004be437d04 0000000000000000 0000000000000000 0000000000000000
GPR12: 0000000000002200 c0000007ffffee00 0000000000000000 0000000000000000
GPR16: 0000000000000000 000000001001cf98 000000001001cf78 000000001001cff0
GPR20: 000000001001d010 000000001001cf58 00007fff92643190 0000000010040530
GPR24: 000000001001d108 000000001001cf48 c0000004be437df0 0000000000000001
GPR28: fffffffffffffff2 0000000000000000 d000000003c22f80 0000000000000000

[69614.978711] NIP [d000000003c200a4] sys_bus_slot+0x8c/0x170 [ltp_tpci]
[69614.978715] LR [d000000003c2006c] sys_bus_slot+0x54/0x170 [ltp_tpci]
[69614.978718] Call Trace:
[69614.978721] [c0000004be437ca0] [d000000003c2006c] 
sys_bus_slot+0x54/0x170 [ltp_tpci] (unreliable)
[69614.978727] [c0000004be437d40] [c0000000008da5bc] 
dev_attr_store+0x3c/0x60
[69614.978732] [c0000004be437d60] [c0000000005fb908] 
sysfs_kf_write+0x68/0x80
[69614.978736] [c0000004be437d80] [c0000000005f9fb4] 
kernfs_fop_write+0x104/0x270
[69614.978741] [c0000004be437dd0] [c0000000004ff624] sys_write+0x134/0x3a0
[69614.978745] [c0000004be437e30] [c00000000000b388] system_call+0x5c/0x70
[69614.978748] Instruction dump:
[69614.978751] ebde8010 80a10064 e87e0000 78bdc622 54bf063e 2fa30000 
419e0014 48000e89
[69614.978757] e8410018 39200000 f93e0000 39200000 <e8690010> 48000dd1 
e8410018 7fe5fb78
[69614.978764] ---[ end trace 414b1270f579351f ]---

=====

Quoted 1 line
#7 lore raw

PATCH 1/1 ltp_tpci.c: fix a null pointer

Yang Xu <[email protected]> wrote:

...

Quoted 10 lines
Quoted 7 lines

I was about to post the same fix on the mailing list - just tested in powerpc arch and it looks good to me.

It is safe to use 0 since function pci_get_bus_and_slot(unsigned int bus, unsigned int devfn) removed on kernel commit <5cf0c37a71da0f3> ("PCI: Remove pci_get_bus_and_slot() function") used to be a wrapper for pci_get_domain_bus_and_slot(0, bus, devfn).

Quoted 3 lines

Sorry for the late reply. Thanks for your review. I think this should be added into commit message, so make this fix more clear.

I helped to add the hint messages and pushed. Thanks for your fix/review.

Signature · 5 lines
Regards,
Li Wang
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20200413/347b917f/attachment.htm>
#8 lore raw

PATCH 1/1 ltp_tpci.c: fix a null pointer

In reply to Yang Xu
Quoted 18 lines · Yang Xu

ps: I want to use a dymaic domain to fix it but failed. If someone know, please tell me. Also, this case failed when merging this patch because test 13 (test_assign_resources) report no space error as below:

[754930.757585] ltp_tpci: test-case 13
[754930.757585] ltp_tpci: assign resources
[754930.757585] ltp_tpci: assign resource #0
[754930.757586] ltp_tpci: name = 0000:00:08.0, flags = 262401, start 0xc140, end 0xc17f
[754930.757586] ltp_tpci: assign resource #1
[754930.757587] ltp_tpci: name = 0000:00:08.0, flags = 262656, start 0xfebd7000, end 0xfebd7fff
[754930.757587] ltp_tpci: assign resource #2
[754930.757587] ltp_tpci: name = 0000:00:08.0, flags = 0, start 0x0, end 0x0
[754930.757588] ltp_tpci: assign resource #3
[754930.757588] ltp_tpci: name = 0000:00:08.0, flags = 0, start 0x0, end 0x0
[754930.757588] ltp_tpci: assign resource #4
[754930.757589] ltp_tpci: name = 0000:00:08.0, flags = 538190348, start 0xfe80c000, end 0xfe80ffff
[754930.757593] virtio-pci 0000:00:08.0: BAR 4: no space for [mem size 0x00004000 64bit pref]
[754930.757594] virtio-pci 0000:00:08.0: BAR 4: failed to assign [mem size 0x00004000 64bit pref

Hi Xu,

Do you mean that the fix patch results in the failure of subtest 13? If so, we may need a better solution.

Thanks,
Xiao Yang