iMX8M PCIe error

cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

iMX8M PCIe error

Jump to solution
2,097 Views
calf
Contributor I

 

Hi,igor:

I have the other issue about iMX8M PCIe application.

The PCIe_1 Bus use the CPU internal PCIe clock. But device(BCM56160) use the external clock from 9FGV0441AKLFT(meet PCIe Gen1/2/3 clock chip). PCIe_1 Bus can work normally. But for a long time(several days or perhaps several weeks), software show a amount of errors on the PCIe_1 access. Maybe that finally will lead the  PCIe_1 Bus link down.

I want to know the root cause for the issue. Do this problem occur because PCIe clock of two sides isn't  the same source?

Do you know the similar issue? Share your experience.

Thanks!

0 Kudos
Reply
1 Solution
2,078 Views
weidong_sun
NXP TechSupport
NXP TechSupport

Hi, Calf,

 

   Don't worry about it.  current PCIe connection can't damage CPU or other parts on board.  Because pcie clock signals doesn't have so large power that it can damage CPU.

so you can ask hardware engineer to look for root cause.

 

Have a nice day!

B.R,

weidong

View solution in original post

0 Kudos
Reply
7 Replies
2,085 Views
weidong_sun
NXP TechSupport
NXP TechSupport

>>If iMX8M's PCIe_A interface use its internal clock, but the PCIe device use the external clock. After working for a long time, whether or not the PCIe_A bus will have a amount of error because of different clock source?

This hardware design is incorrect, which means that the PCIe controller and the PCIE device use different clock sources, which cannot guarantee data synchronization. Therefore, you will find a lot of errors in PCIe transmission.

 

Have a nice day!

weidong

0 Kudos
Reply
2,082 Views
calf
Contributor I

Hi, Weidong:

Thank for your reply. That verfiy our test about PCIe_A bus error.

In addition, I want to know whether or not damage the PCIe_A bus hardware under data asynchronization.  Because our CPU(iMX8M) have five pcs damaged and all CPUs have a PCIe_A bus destroyed. PCIe_A_TX has no signal after startup. PCIe_A clock use the chip's internal clock.

Please give me idea or suggestion.

BRs

calf

0 Kudos
Reply
2,079 Views
weidong_sun
NXP TechSupport
NXP TechSupport

Hi, Calf,

 

   Don't worry about it.  current PCIe connection can't damage CPU or other parts on board.  Because pcie clock signals doesn't have so large power that it can damage CPU.

so you can ask hardware engineer to look for root cause.

 

Have a nice day!

B.R,

weidong

0 Kudos
Reply
2,054 Views
calf
Contributor I

Hi, Weidong;

When PCIe clock is  asynchronized between CPU and device, that lead CPU is overheating because PCIe access failure lead to retry again and again. Can CPU's PCIe_TX pins be damaged after working for a long time(maybe several weeks or months)?

 

BRs

calf

0 Kudos
Reply
2,076 Views
calf
Contributor I

Hi, Weidong;

Thank very much for your great support!

 

BRs

calf

0 Kudos
Reply
2,095 Views
weidong_sun
NXP TechSupport
NXP TechSupport

Hi,

Just drew a diagram for you. see below, please!

pcie-clock.png

Hope it is helpful to you.

Have a nice day!

B.R,

weidong

0 Kudos
Reply
2,091 Views
calf
Contributor I

Hi, Weidong:

If iMX8M's PCIe_A interface use its internal clock, but the PCIe device use the external clock. After working for a long time, whether or not the PCIe_A bus will have a amount of error because of different clock source?

Do you have the similar issue on other customers?

BRs

calf

0 Kudos
Reply