Skip to contentSkip to conversation

ltp_tpci.c: Add release operation before allocation

PATCH New Wang Xin 3 participants 10 messages Apr 1, 2021 – Jul 26, 2021

lore
Share & tools
lore

Wang Xin

author
#0 lore raw

PATCH 1/1 ltp_tpci.c: Add release operation before allocation

When we run the test case, the following results will be obtained:

test_pci  283  TPASS  :  PCI bus 7d slot 00 : Test-case '11'
test_pci  284  TFAIL  :  tpci.c:74: PCI bus 7d slot 00 : Test-case '12'
...
test_pci  300  TFAIL  :  tpci.c:74: PCI bus 7d slot 01 : Test-case '12'
test_pci  301  TPASS  :  PCI bus 7d slot 01 : Test-case '13'

The analysis is that the space allocated by the bios is insufficient. The solution to this problem can be in add pci_release_resource(dev, i) before system resources are reallocated.Because the resources have been allocated when the system is initialized.If it is redistributed, it should be released and then allocated.

---
 testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c | 1 +
 1 file changed, 1 insertion(+)
testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c +1 -0
index e29821dda..5b0896093 100644
@@ -439,6 +439,7 @@ static int test_assign_resources(void)
 		if (r->flags & IORESOURCE_MEM &&
 			r->flags & IORESOURCE_PREFETCH) {
+			pci_release_resource(dev, i);
 			ret = pci_assign_resource(dev, i);
 			prk_info("assign resource to '%d', ret '%d'", i, ret);
 			rc |= (ret < 0 && ret != -EBUSY) ? TFAIL : TPASS;

2.17.1

#1 lore raw

PATCH 1/1 ltp_tpci.c: Add release operation before allocation

Hi Xin,

Quoted 7 lines

When we run the test case, the following results will be obtained:

test_pci  283  TPASS  :  PCI bus 7d slot 00 : Test-case '11'
test_pci  284  TFAIL  :  tpci.c:74: PCI bus 7d slot 00 : Test-case '12'
...
test_pci  300  TFAIL  :  tpci.c:74: PCI bus 7d slot 01 : Test-case '12'
test_pci  301  TPASS  :  PCI bus 7d slot 01 : Test-case '13'
Quoted 5 lines

The analysis is that the space allocated by the bios is insufficient. The solution to this problem can be in add pci_release_resource(dev, i) before system resources are reallocated.Because the resources have been allocated when the system is initialized.If it is redistributed, it should be released and then allocated.

I wonder if this is the same issue as the one described by Krzysztof in his patch:
https://patchwork.ozlabs.org/project/ltp/patch/[email protected]/

Could you please share what HW and kernel you use and post dmesg?

Kind regards,
Petr

Quoted 17 lines
---
 testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c | 1 +
 1 file changed, 1 insertion(+)
testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c +1 -0
index e29821dda..5b0896093 100644
@@ -439,6 +439,7 @@ static int test_assign_resources(void)
 		if (r->flags & IORESOURCE_MEM &&
 			r->flags & IORESOURCE_PREFETCH) {
+			pci_release_resource(dev, i);
 			ret = pci_assign_resource(dev, i);
 			prk_info("assign resource to '%d', ret '%d'", i, ret);
 			rc |= (ret < 0 && ret != -EBUSY) ? TFAIL : TPASS;

2.17.1

#2 lore raw

PATCH 1/1 ltp_tpci.c: Add release operation before allocation

Quoted 15 lines · Petr Vorel

Hi Xin,

Quoted 7 lines

When we run the test case, the following results will be obtained:

test_pci  283  TPASS  :  PCI bus 7d slot 00 : Test-case '11'
test_pci  284  TFAIL  :  tpci.c:74: PCI bus 7d slot 00 : Test-case '12'
...
test_pci  300  TFAIL  :  tpci.c:74: PCI bus 7d slot 01 : Test-case '12'
test_pci  301  TPASS  :  PCI bus 7d slot 01 : Test-case '13'
Quoted 5 lines

The analysis is that the space allocated by the bios is insufficient. The solution to this problem can be in add pci_release_resource(dev, i) before system resources are reallocated.Because the resources have been allocated when the system is initialized.If it is redistributed, it should be released and then allocated.

Quoted 5 lines

I wonder if this is the same issue as the one described by Krzysztof in his patch:
https://patchwork.ozlabs.org/project/ltp/patch/[email protected]/

Could you please share what HW and kernel you use and post dmesg?

This solves my problem. It seems that could be the root cause - early configuration allocated too small resource? It might be also some specific BIOS issue (wrong resource allocated?) because in case of resource assignment failure, the kernel should try to get the original FW address from BIOS (pcibios_retrieve_fw_addr()) and this apparently returns NULL translated to -ENOMEM.

I am fine with going with this patch instead of mine.

Best regards,
Krzysztof

#3 lore raw

PATCH 1/1 ltp_tpci.c: Add release operation before allocation

Quoted 22 lines · Krzysztof Kozlowski
Quoted 15 lines · Petr Vorel

Hi Xin,

Quoted 7 lines

When we run the test case, the following results will be obtained:

test_pci  283  TPASS  :  PCI bus 7d slot 00 : Test-case '11'
test_pci  284  TFAIL  :  tpci.c:74: PCI bus 7d slot 00 : Test-case '12'
...
test_pci  300  TFAIL  :  tpci.c:74: PCI bus 7d slot 01 : Test-case '12'
test_pci  301  TPASS  :  PCI bus 7d slot 01 : Test-case '13'
Quoted 5 lines

The analysis is that the space allocated by the bios is insufficient. The solution to this problem can be in add pci_release_resource(dev, i) before system resources are reallocated.Because the resources have been allocated when the system is initialized.If it is redistributed, it should be released and then allocated.

Quoted 5 lines

I wonder if this is the same issue as the one described by Krzysztof in his patch:
https://patchwork.ozlabs.org/project/ltp/patch/[email protected]/

Could you please share what HW and kernel you use and post dmesg?

Quoted 9 lines

This solves my problem. It seems that could be the root cause - early configuration allocated too small resource? It might be also some specific BIOS issue (wrong resource allocated?) because in case of resource assignment failure, the kernel should try to get the original FW address from BIOS (pcibios_retrieve_fw_addr()) and this apparently returns NULL translated to -ENOMEM.

I am fine with going with this patch instead of mine.

Is there anything stopping Xin's patch from being applied? LGTM:

Best regards,
Krzysztof

#4 lore raw

PATCH 1/1 ltp_tpci.c: Add release operation before allocation

Quoted 30 lines · Krzysztof Kozlowski
Quoted 29 lines · Krzysztof Kozlowski
Quoted 15 lines · Petr Vorel

Hi Xin,

Quoted 7 lines

When we run the test case, the following results will be obtained:

test_pci  283  TPASS  :  PCI bus 7d slot 00 : Test-case '11'
test_pci  284  TFAIL  :  tpci.c:74: PCI bus 7d slot 00 : Test-case '12'
...
test_pci  300  TFAIL  :  tpci.c:74: PCI bus 7d slot 01 : Test-case '12'
test_pci  301  TPASS  :  PCI bus 7d slot 01 : Test-case '13'
Quoted 5 lines

The analysis is that the space allocated by the bios is insufficient. The solution to this problem can be in add pci_release_resource(dev, i) before system resources are reallocated.Because the resources have been allocated when the system is initialized.If it is redistributed, it should be released and then allocated.

Quoted 5 lines

I wonder if this is the same issue as the one described by Krzysztof in his patch:
https://patchwork.ozlabs.org/project/ltp/patch/[email protected]/

Could you please share what HW and kernel you use and post dmesg?

This solves my problem. It seems that could be the root cause - early configuration allocated too small resource? It might be also some specific BIOS issue (wrong resource allocated?) because in case of resource assignment failure, the kernel should try to get the original FW address from BIOS (pcibios_retrieve_fw_addr()) and this apparently returns NULL translated to -ENOMEM.

Quoted 7 lines
Quoted 2 lines

I am fine with going with this patch instead of mine.

Is there anything stopping Xin's patch from being applied? LGTM:

Oh, I see what's wrong with the patch - it's corrupted, not possible to apply.

Best regards,
Krzysztof

#5 lore raw

PATCH 1/1 ltp_tpci.c: Add release operation before allocation

Hi all,

Quoted 31 lines · Krzysztof Kozlowski
Quoted 29 lines · Krzysztof Kozlowski
Quoted 28 lines · Petr Vorel
Quoted 15 lines

Hi Xin,

Quoted 7 lines

When we run the test case, the following results will be obtained:

test_pci  283  TPASS  :  PCI bus 7d slot 00 : Test-case '11'
test_pci  284  TFAIL  :  tpci.c:74: PCI bus 7d slot 00 : Test-case '12'
...
test_pci  300  TFAIL  :  tpci.c:74: PCI bus 7d slot 01 : Test-case '12'
test_pci  301  TPASS  :  PCI bus 7d slot 01 : Test-case '13'
Quoted 5 lines

The analysis is that the space allocated by the bios is insufficient. The solution to this problem can be in add pci_release_resource(dev, i) before system resources are reallocated.Because the resources have been allocated when the system is initialized.If it is redistributed, it should be released and then allocated.

Quoted 5 lines

I wonder if this is the same issue as the one described by Krzysztof in his patch:
https://patchwork.ozlabs.org/project/ltp/patch/[email protected]/

Could you please share what HW and kernel you use and post dmesg?

This solves my problem. It seems that could be the root cause - early configuration allocated too small resource? It might be also some specific BIOS issue (wrong resource allocated?) because in case of resource assignment failure, the kernel should try to get the original FW address from BIOS (pcibios_retrieve_fw_addr()) and this apparently returns NULL translated to -ENOMEM.

Quoted 4 lines
Quoted 4 lines
Quoted 1 line

I am fine with going with this patch instead of mine.

Is there anything stopping Xin's patch from being applied? LGTM:

Lack of time :)

Quoted 1 line
Quoted 1 line
Quoted 2 lines

Oh, I see what's wrong with the patch - it's corrupted, not possible to apply.

Trivial fix, I'll merge it shortly.

Kind regards,
Petr

Quoted 2 lines

Best regards,
Krzysztof

#7 lore raw

PATCH 1/1 ltp_tpci.c: Add release operation before allocation

Quoted 7 lines · Petr Vorel

Hi Krzysztof

...

Quoted 1 line
Quoted 1 line

Is there anything stopping Xin's patch from being applied? LGTM:

BTW does it solve your issue "tpci: accept ENOMEM resource failure with virtio-pci" [1] ?

Yes, it solves. My patch can be dropped. Thanks!

Best regards,
Krzysztof

#8 lore raw

PATCH 1/1 ltp_tpci.c: Add release operation before allocation

Quoted 10 lines · Petr Vorel
Quoted 7 lines

Hi Krzysztof

...

Quoted 1 line
Quoted 1 line

Is there anything stopping Xin's patch from being applied? LGTM:

BTW does it solve your issue "tpci: accept ENOMEM resource failure with virtio-pci" [1] ?

Yes, it solves. My patch can be dropped. Thanks!

Good, thanks for confirmation.

Merged!

Kind regards,
Petr

Quoted 2 lines

Best regards,
Krzysztof

#9 lore raw
In reply to Petr Vorel

Hi Petr,

The following is the kernel version information of the test case. The hardware information is in the attachment.

# uname -a
Linux hulk-168 4.19.195+ #1 SMP Tue Jun 29 09:42:18 EDT 2021 aarch64 aarch64 aarch64 GNU/Linux
# lscpu 
Architecture:        aarch64
Byte Order:          Little Endian
CPU(s):              128
On-line CPU(s) list: 0-127
Thread(s) per core:  1
Core(s) per socket:  64
Socket(s):           2
NUMA node(s):        4
Vendor ID:           0x48
Model:               0
Stepping:            0x1
CPU max MHz:         2600.0000
CPU min MHz:         200.0000
BogoMIPS:            200.00
L1d cache:           64K
L1i cache:           64K
L2 cache:            512K
L3 cache:            65536K
NUMA node0 CPU(s):   0-31
NUMA node1 CPU(s):   32-63
NUMA node2 CPU(s):   64-95
NUMA node3 CPU(s):   96-127
Flags:               fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma dcpop asimddp asimdfhm

Kind regards,
Xin

-----????-----
???: Petr Vorel [mailto:[email protected]]
????: 2021?6?29? 4:30
???: wangxin (CQ) <[email protected]>
??: [email protected]; Krzysztof Kozlowski <[email protected]> ??: Re: [LTP] [PATCH] ltp_tpci.c: Add release operation before allocation

Hi Xin,

Quoted 7 lines

When we run the test case, the following results will be obtained:

test_pci  283  TPASS  :  PCI bus 7d slot 00 : Test-case '11'
test_pci  284  TFAIL  :  tpci.c:74: PCI bus 7d slot 00 : Test-case '12'
...
test_pci  300  TFAIL  :  tpci.c:74: PCI bus 7d slot 01 : Test-case '12'
test_pci  301  TPASS  :  PCI bus 7d slot 01 : Test-case '13'
Quoted 5 lines

The analysis is that the space allocated by the bios is insufficient. The solution to this problem can be in add pci_release_resource(dev,
i) before system resources are reallocated.Because the resources have been allocated when the system is initialized.If it is redistributed, it should be released and then allocated.

I wonder if this is the same issue as the one described by Krzysztof in his patch:
https://patchwork.ozlabs.org/project/ltp/patch/[email protected]/

Could you please share what HW and kernel you use and post dmesg?

Kind regards,
Petr

Quoted 18 lines
---
 testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c | 1 +
 1 file changed, 1 insertion(+)
(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 e29821dda..5b0896093 100644

testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c +1 -0
@@ -439,6 +439,7 @@ static int test_assign_resources(void)
 		if (r->flags & IORESOURCE_MEM &&
 			r->flags & IORESOURCE_PREFETCH) {
+			pci_release_resource(dev, i);
 			ret = pci_assign_resource(dev, i);
 			prk_info("assign resource to '%d', ret '%d'", i, ret);
 			rc |= (ret < 0 && ret != -EBUSY) ? TFAIL : TPASS;
Quoted 1 line

2.17.1

-------------- next part --------------
A non-text attachment was scrubbed...
Name: dmesg.log
Type: application/octet-stream
Size: 135634 bytes
Desc: dmesg.log
URL: <http://lists.linux.it/pipermail/ltp/attachments/20210630/6e9839e7/attachment-0001.obj>