A method for resolving data mistransmission in NVME RAID groups when VMD is disabled.

Abstract

The present application relates to a kind of methods for solving NVME RAID group data error transmission when VMD is closed, belong to hardware platform design field.The data link channel built by the data link channel using LPC protocol from PCH to BMC, IIC protocol from BMC to CPLD is used to transmit VMD information data;The present application avoids the data disorder caused by the unstable PCH drive of platform during cold start or reset by judging the VMD state, avoids the error indication of the LED indicator light of NVME hard disk.The processing flow of the present application can solve the two problems mentioned above, avoid the data error transmission problem caused by the disorder of PCH drive during Boot phase.Making even not open VMD, hard disk state display can also normally display Present state and Active state under BIOS, and the state can be correctly maintained.

Description

Technical Field

[0001] This invention belongs to the field of hardware platform design, specifically relating to a method for resolving data mistransmission in NVME RAID groups when VMD is disabled. Background Technology

[0002] Servers in enterprises and institutions are typically used for database services, video services, and file services. These applications usually have stringent requirements for server hard drive capacity, read / write speed, and data security. To meet these three requirements simultaneously, enterprises and institutions often use large-capacity NVMe hard drive RAID arrays.

[0003] Currently, the x86 platform holds a significant market share in the server market, while the Intel platform holds an absolute advantage. Intel's Purley, Whitely, and EagleStream platforms all support RAID 0, RAID 1, and RAID 3 functionality for NVMe hard drives under BIOS setup. After NVMe hard drives are configured into a RAID group, the hard drive status within the RAID group involves several aspects, including data read / write (Activity), hard drive location (Locate), RAID group corruption (Fault), and RAID reconstruction (Rebuild). Intel allows RAID group configuration by enabling the VMD function under BIOS setup. With VMD enabled, the hard drive status data sent by the PCH in the RAID group is normal. If the user enters the OS without enabling VMD under BIOS setup, the PCH will default to sending information unrelated to the current hard drive status (in the absence of RAID, the PCH sends Fault and Rebuild data). If the user avoids corrupted data by blocking the PCH data channel, then when VMD is enabled, the hard drive status data in the normal RAID group cannot be received by the peripheral management device (CPLD).

[0004] In Intel server platforms, NVMe hard drive status data is sent from the PCH to the CPLD on the hard drive backplane via the VPP protocol. Existing Intel platform servers have two issues.

[0005] 1: If the VMD option is not enabled in the BIOS setup, the PCH will send Fault and Locate data by default after entering the OS.

[0006] 2: After the PCH issues Fault or Locate data, it cannot issue a clear data command to correct the erroneous data status itself.

[0007] In Intel server platforms, the CPLD parses hard drive status data from the VPP protocol and controls LEDs to display it. If the aforementioned issue is not resolved, the displayed hard drive status light status will not correspond to the actual working status of the hard drive.

[0008] Server maintenance personnel typically determine the current server status by observing various status indicator lights and sensor information displayed on the web management interface. If there are discrepancies between the hard drive status indicator lights and the actual hard drive status information displayed on the system, it can easily lead to misjudgments of the server's operating status by maintenance personnel, affecting maintenance work and business operations. Summary of the Invention

[0009] (a) Technical problems to be solved

[0010] The technical problem to be solved by this invention is how to provide a method for resolving the mistransmission of NVME RAID group data when VMD is turned off, so as to solve the problem that the status displayed by the hard drive status light does not correspond to the actual working status of the hard drive.

[0011] (II) Technical Solution

[0012] To address the aforementioned technical problems, this invention proposes a method for resolving data mistransmission in NVME RAID groups when VMD is disabled. This method is implemented through the cooperation of the Platform Controller Center (PCH) and Baseboard Management Controller (BMC) on the server motherboard and the CPLD on the hard drive backplane. The PCH and BMC on the server motherboard communicate via the LPC protocol, and the BMC and CPLD communicate via the IIC protocol. When the NVME hard drive group is in RAID mode, the PCH sends hard drive data to the CPLD on the backplane via the VPP protocol. The PCH sends a Boot Complete signal to the CPLD via GPIO. The CPLD obtains hard drive read / write status data through Pin 11 of the U.2 hard drive connector and obtains hard drive presence data through Pin 10.

[0013] When the server performs a cold start or restart, the BIOS begins to boot the CPU. The PCH sends the BootComplete signal to the CPLD via GPIO. The Complete signal is active low and is pulled up by default.

[0014] If the CPLD detects that the Boot Complete signal is high, it assumes that the BIOS boot process has not been completed. The CPLD actively blocks the hard drive data sent by the PCH through the VPP. At this time, the CPLD only parses the data of Pin 10 and Pin 11 of the hard drive connector and controls the corresponding LEDs.

[0015] If the CPLD detects that the Boot Complete signal is low, the CPLD will remove the data restriction on the VPP and make a second judgment based on the VMD information sent by the BMC.

[0016] If VMD is enabled, the BIOS can set up RAID under setup or OS. At this time, CPLD can parse the data in VPP and integrate the data of Pin10 and Pin11 to control the on / off state of the corresponding LEDs.

[0017] If the VMD function is disabled, the BIOS cannot set up RAID under setup or OS. In this case, the CPLD only parses the data of Pin 10 and Pin 11 of the hard drive connector and controls the corresponding LEDs.

[0018] Furthermore, this method can be applied to Intel's Purley, Whitely, or EagleStream platforms.

[0019] Furthermore, when the NVMe hard drive is in RAID state, the hard drive status includes: hard drive present, hard drive absent, hard drive read / write active, hard drive rebuild in RAID group, hard drive location, and RAID group corruption fault.

[0020] Furthermore, when the NVMe hard drive is in a non-RAID state, the hard drive status includes: hard drive present, hard drive absent, and hard drive read / write active.

[0021] Furthermore, when the hard drive is reading or writing data, it will send a square wave signal to the CPLD through Pin 11 of the hard drive connector.

[0022] Furthermore, Pin 10 is the hard drive presence signal pin. This signal is externally pulled up to a high level by the hardware and connected to the CPLD via an internal trace. When the hard drive is inserted into the U.2 hard drive connector, Pin 10 is shorted to the GND inside the hard drive, and the signal is pulled down to a low level.

[0023] Furthermore, VMD is used on Intel platforms to control whether RAID functionality is enabled or disabled. If VMD is enabled, the BIOS can set up RAID in setup mode or under the OS; if VMD is disabled, the BIOS cannot set up RAID in setup mode or under the OS.

[0024] Furthermore, VMD information is sent by the BIOS to the BMC via the LPC protocol.

[0025] Furthermore, the CPLD only parses the data of Pin 10 and Pin 11 of the hard drive connector and controls the corresponding LEDs, including: hard drive read / write indicator and hard drive present indicator.

[0026] Furthermore, the CPLD allows parsing data in the VPP and integrating the data from Pin10 and Pin11 to control the on / off state of the corresponding LEDs, including: when a hard drive in the RAID group is located, the CPLD controls the LED of the corresponding hard drive to flash in the locate state; when a hard drive in the RAID group is removed, the hard drive removed from the slot will flash in the faut state; when a hard drive in the RAID group is rebuilt, it will flash in the rebuild state.

[0027] (III) Beneficial Effects

[0028] This invention proposes a method to solve the problem of mistransmission of NVME RAID group data when VMD is shut down. The data link channel established by this invention using LPC protocol from PCH to BMC and IIC protocol from BMC to CPLD is used to transmit VMD information data.

[0029] This invention avoids data corruption caused by PCH driver instability during cold starts or resets by judging the VMD status, and also avoids erroneous indication of the LED indicator lights on NVMe hard drives.

[0030] The processing flow of this invention solves the two problems mentioned above, avoiding data mistransmission caused by driver disorder under the PCH during the Boot stage. This ensures that even without VMD enabled in the BIOS, the hard drive status display can correctly show the Present and Active states, and the states can be correctly maintained. Attached Figure Description

[0031] Figure 1 This is a system architecture diagram of the present invention used to solve the problem of data mistransmission in NVME RAID groups when VMD is turned off;

[0032] Figure 2 This is a schematic diagram of the hard drive status information transmission path;

[0033] Figure 3 The logic diagram for Boot Complete judgment;

[0034] Figure 4 The logic diagram for VMD judgment. Detailed Implementation

[0035] To make the objectives, contents, and advantages of the present invention clearer, the specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples.

[0036] Definitions:

[0037] PCH: Platform Controller Hub (Intel's integrated southbridge)

[0038] LPC: Low Pin Count Bus, Low Speed ​​Device Bus

[0039] VPP: Virtual Port Pin (Virtual Device Platform Interface)

[0040] BMC: Baseboard Management Controller

[0041] VMD: Volume Management Device

[0042] This invention relates to addressing the technical deficiencies of existing Intel x86 platforms by improving compatibility issues through onboard hardware CPLD, BMC, and PCH, and via software-level data interaction.

[0043] This invention describes a solution based on the Purley platform (applicable to Whitley and Eagle Stream). The design is implemented using the PCH and BMC on the server motherboard and the CPLD on the hard drive backplane.

[0044] like Figure 1 As shown, the communication link of the present invention is as follows:

[0045] 1: The server motherboard PCH and BMC communicate via the LPC protocol; LPC is the communication protocol between the BIOS and BMC, in which the BIOS controls the PCH to send information to the BMC;

[0046] 2: The BMC and CPLD communicate via the IIC protocol.

[0047] 3: When using NVMe hard drives in RAID configuration, the PCH sends hard drive data to the backplane CPLD via the VPP protocol;

[0048] 4: The PCH sends a Boot Complete signal to the CPLD via GPIO;

[0049] 5. The status of NVMe hard drives differs significantly depending on whether they are in RAID or non-RAID mode. When in RAID mode, the drive status includes Present, Disappear, Active (read / write), Rebuild (RAID group drive reconstruction), Locate, and Fault (RAID group corruption). When in non-RAID mode, the drive status includes Present, Disappear, and Active (read / write).

[0050] 5: The CPLD obtains hard drive read / write status data through Pin 11 of the U.2 hard drive connector (connector specifications comply with SFF-8639 standard), and obtains hard drive presence data through Pin 10. This is unrelated to whether the NVMe hard drive is in RAID or non-RAID state. The Active state and Present information are determined by the hard drive itself.

[0051] The channels for obtaining hard drive status information are as follows: Figure 2 As shown. Pins 11 and 10 of the U.2 hard drive connector are connected to the CPLD via internal signal lines on the backplane. When the hard drive is reading or writing data, it emits a square wave signal, which is sent to the CPLD through Pin 11 of the hard drive connector. Pin 10 is the hard drive presence signal pin. This signal is externally pulled up to a high level by hardware and simultaneously connected to the CPLD via an internal trace. When the hard drive is inserted into the U.2 hard drive connector, Pin 10 is shorted to the hard drive's internal GND, and the signal is pulled down to a low level.

[0052] like Figure 3 As shown, when the server performs a cold boot or restart, the BIOS begins booting the CPU. The PCH sends a Boot Complete signal (pull-up by default, active low) to the CPLD via GPIO. If the CPLD detects the Boot Complete signal as high, it considers the BIOS boot process incomplete and actively blocks hard drive data sent by the PCH through the VPP. If the CPLD detects the Boot Complete signal as low, it removes the data restriction on the VPP and performs a secondary judgment based on the VMD information sent by the BMC. VMD, short for Volume Device Management, is used in Intel platforms to control whether RAID functionality is enabled or disabled. If VMD is enabled, the BIOS can create RAID arrays during setup or under the operating system; if VMD is disabled, the BIOS cannot create RAID arrays during setup or under the operating system. In this invention, the VMD information indicates whether RAID arrays are allowed.

[0053] VMD information is sent by the BIOS to the BMC via the LPC protocol, which stands for Low Speed ​​Device Protocol. If the hard drives are configured in RAID, the PCH will send correct data to the CPLD via VPP. If the hard drives are not configured in RAID, the VMD information will show that the hard drives are not configured in RAID, but the PCH will still send incorrect hard drive data via VPP in this case.

[0054] like Figure 4 As shown, after the CPLD receives the Boot Complete signal sent by the BIOS, if the Boot Complete signal is high, the CPLD only parses the data of Pin 10 and Pin 11 of the hard drive connector and controls the corresponding LEDs. At this time, Pin 10 indicates the hard drive read / write information indicator, which flashes when the hard drive is read or written, and Pin 11 indicates the hard drive presence information indicator.

[0055] If Boot Complete is low, the CPLD checks the VMD information sent by the BMC via IIC. If VMD is enabled, the CPLD allows parsing the data in the VPP and integrating the data from Pin 10 and Pin 11, controlling the on / off state of the corresponding LEDs. If VMD is enabled, the hard drives can be configured into RAID. At this time, the hard drive indicator lights will illuminate and deactivate according to the specific hard drive status. For example, when locating a hard drive in the RAID group, the CPLD controls the corresponding hard drive's LED to flash in the "locate" state. When a hard drive in the RAID group is removed, the removed hard drive will flash in the "fautl" state. When a hard drive in the RAID group is being rebuilt, it will flash in the "rebuild" state.

[0056] If VMD is not enabled, CPLD will only parse the data in Pin 10 and Pin 11 of the hard drive connector. In this case, when VMD is not enabled, the hard drive status indicator light can only flash in an active (hard drive read / write) mode, or when the hard drive is not being read / written, the status indicator light will be constantly on when the hard drive is in place, and off when the hard drive is not in place.

[0057] This processing flow resolves the two issues mentioned above, preventing data mistransmission caused by driver corruption during the Boot phase PCH. It ensures that even without VMD enabled in the BIOS, the hard drive status display correctly shows Present and Active states, and these states are maintained correctly.

[0058] The present invention uses the LPC protocol to establish a data link channel from PCH to BMC and the IIC protocol to establish a data link channel from BMC to CPLD to transmit VMD information data.

[0059] This invention avoids data corruption caused by PCH driver instability during cold starts or resets by judging the VMD status, and also avoids erroneous indication of the LED indicator lights on NVMe hard drives.

[0060] The processing flow of this invention solves the two problems mentioned above, avoiding data mistransmission caused by driver disorder under the PCH during the Boot stage. This ensures that even without VMD enabled in the BIOS, the hard drive status display can correctly show the Present and Active states, and the states can be correctly maintained.

[0061] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for resolving data mistransmission in NVME RAID groups when VMD is disabled, characterized in that, This method is implemented by the Platform Controller Center (PCH) and Baseboard Management Controller (BMC) on the server motherboard, and the CPLD on the hard drive backplane. The PCH and BMC on the server motherboard communicate with each other via the LPC protocol, and the BMC and CPLD communicate with each other via the IIC protocol. When the NVMe hard drive is in RAID configuration, the PCH sends hard drive data to the backplane CPLD via the VPP protocol. The PCH sends a BootComplete signal to the CPLD via GPIO. The CPLD obtains hard drive read / write status data through Pin 11 of the U.2 hard drive connector and obtains hard drive presence data through Pin 10. When the server performs a cold start or restart, the BIOS begins to boot the CPU. The PCH sends a Boot Complete signal to the CPLD via GPIO. The Complete signal is active low and is pulled up by default. If the CPLD detects that the Boot Complete signal is high, it assumes that the BIOS boot process has not been completed. The CPLD actively blocks the hard drive data sent by the PCH through the VPP. At this time, the CPLD only parses the data of Pin 10 and Pin 11 of the hard drive connector and controls the corresponding LEDs. If the CPLD detects that the Boot Complete signal is low, the CPLD will remove the data restriction on the VPP and make a second judgment based on the VMD information sent by the BMC. If VMD is enabled, the BIOS can set up RAID under setup or OS. At this time, CPLD can parse the data in VPP and integrate the data of Pin10 and Pin11 to control the on / off state of the corresponding LEDs. If the VMD function is disabled, the BIOS cannot set up RAID under setup or OS. In this case, the CPLD only parses the data of Pin 10 and Pin 11 of the hard drive connector and controls the corresponding LEDs.

2. The method for resolving data mistransmission in NVME RAID groups when VMD is disabled, as described in claim 1, is characterized in that... This method can be applied to Intel's Purley, Whitely, or EagleStream platforms.

3. The method for resolving data mistransmission in NVME RAID groups when VMD is disabled, as described in claim 1, is characterized in that... When an NVMe hard drive is in RAID mode, the hard drive status includes: present, absent, active, rebuild, location, and fault.

4. The method for resolving data mistransmission in NVME RAID groups when VMD is disabled, as described in claim 1, is characterized in that... When an NVMe hard drive is in a non-RAID state, the hard drive status includes: hard drive present, hard drive absent, and hard drive read / write active.

5. The method for resolving data mistransmission in NVME RAID groups when VMD is disabled, as described in claim 1, characterized in that... When the hard drive is reading or writing data, it sends a square wave signal to the CPLD through Pin 11 of the hard drive connector.

6. The method for resolving data mistransmission in NVME RAID groups when VMD is disabled, as described in claim 1, is characterized in that... Pin10 is the hard drive presence signal pin. This signal is externally pulled up to a high level by the hardware and connected to the CPLD via an internal trace. When the hard drive is inserted into the U.2 hard drive connector, Pin10 is shorted to the internal GND of the hard drive, and the signal is pulled down to a low level.

7. The method for resolving data mistransmission in an NVME RAID group when VMD is disabled, as described in any one of claims 1-6, characterized in that, In Intel platforms, VMD is used to control whether RAID functionality is enabled or disabled. If VMD is enabled, the BIOS can set up RAID in setup mode or under the OS. If VMD is disabled, the BIOS cannot set up RAID in setup mode or under the OS.

8. The method for resolving data mistransmission in an NVME RAID group when VMD is disabled, as described in claim 7, is characterized in that... VMD information is sent by the BIOS to the BMC via the LPC protocol.

9. The method for resolving data mistransmission in an NVME RAID group when VMD is disabled, as described in claim 7, is characterized in that... The CPLD only parses the data of Pin 10 and Pin 11 of the hard drive connector and controls the corresponding LEDs, including: hard drive read / write indicator and hard drive present indicator.

10. The method for resolving data mistransmission in an NVME RAID group when VMD is disabled, as described in claim 7, characterized in that... The CPLD allows parsing data in the VPP and integrating the data from Pin 10 and Pin 11 to control the on / off state of the corresponding LEDs, including: when a hard drive in the RAID group is located, the CPLD controls the LED of the corresponding hard drive to flash in the locate state; when a hard drive in the RAID group is removed, the hard drive removed from the slot will flash in the faut state; when a hard drive in the RAID group is rebuilt, it will flash in the rebuild state.

Images