PCIDriverKit

RSS for tag

Develop device drivers for Peripheral Component Interconnect accessories.

Posts under PCIDriverKit tag

14 Posts
Sort by:

Post

Replies

Boosts

Views

Activity

Issue with IOServiceOpen
I have an app that is used to control features of a device with a driverkit driver. I am having trouble creating a connection a certain device. The Sample code from "Communicating between a DriverKit extension and a client app. The sample code shows: ret = IOServiceGetMatchingServices(kIOMainPortDefault, IOServiceNameMatching(kDextIdentifier), &iterator); I cannot use kDextIdentifier but need to find a service with a certain BSD Name. So in this case I try: ret = IOServiceGetMatchingServices(kIOMainPortDefault, IOBSDNameMatching(kIOMasterPortDefault, NULL, interface), &iterator); In each case the call completes correctly, and we get an iterator. I can also use IOServiceGetMatchingService with IOBSDNameMatching, and that completes correctly as well. However when I attempt IOServiceOpen with the first case, the connection is created correctly. However, I have four of these in the machine, and I need to select the service and subsequently connection for a certain BSD name. When I attempt the IOServiceOpen with the second/third calls, the IOServiceOpen call fails with error 0x2c7 which is unsupported. Is there an entitlement I need to make this work?
1
0
109
6d
How to enable PCIDriverKit Bus Leader? (and Memory Space enable?)
I am porting a working kernel extension IOKit driver to a DriverKit system extension. Our device is a PCI device accessed through Thunderbolt. The change from IOPCIFamily to PCIDriverKit has some differences in approach, though. Namely, in IOKit / IOPCIFamily, this was the correct way to become Bus Leader: mPCIDevice->setBusLeadEnable(true); // setBusMasterEnable(..) deprecated in OS 12.4 but now, PCIDriverKit's IOPCIDevice does not have that function. Instead I am doing the following: // Set Bus Leader and Memory Space enable uint16_t commandRegister = 0; ivars->mPCIDevice->ConfigurationRead16(kIOPCIConfigurationOffsetCommand, &commandRegister); commandRegister |= (kIOPCICommandBusLead | kIOPCICommandMemorySpace); ivars->mPCIDevice->ConfigurationWrite16(kIOPCIConfigurationOffsetCommand, commandRegister); But I am not convinced this is working (I am still experiencing unexpected errors when attempting to DMA from our device, using the same steps that work for the kernel extension). The only hint I can find in the online documentation is here, which reads: Note The endpoint driver is responsible for enabling the Memory Space Enable and Bus Master Enable settings each time it configures the PCI device. When a crash occurs, or when the system unloads your driver, the system disables these features. ...but that does not state directly how to enable bus leader status. What is the "PCIDriverKit approved" way to become bus leader? Is there a way to verify/confirm that a device is bus leader? (This would be helpful to prove that bus leadership is not the issue for DMA errors, as well as to confirm that bus leadership was granted). Thanks in advance!
0
0
76
1w
Dext signing issue on Sequoia Beta
I am developing a PCIDriverKit dext, and testing on Sequoia Beta (Version 15.0 Beta, 24A5298h). Both the dext and the "owning" application build on Xcode 16.0 beta 4. I can run the owning application and register the dext. When the OS attempts to load the dext, though, code signing validation errors occur: 2024-07-30 15:54:02.386 Df kernel[0:ae6a] Driver com.company.Dext-Loader.dext has crashed 0 time(s) 2024-07-30 15:54:02.386 Df kernel[0:ae6a] DK: Dext_Loader_Driver-0x100001464 waiting for server com.company.Dext-Loader.dext-100001464 2024-07-30 15:54:02.388 Df kernelmanagerd[112:abb5] Found 1 dexts with bundle identifier com.company.Dext-Loader.dext 2024-07-30 15:54:02.388 Df kernelmanagerd[112:abb5] Using unique id a0cf49ca3ea45f5d54a3e8644e2dde6b0e8666c649c1e9513ca4166919038b53 to pick dext matching bundle identifier com.company.Dext-Loader.dext 2024-07-30 15:54:02.388 Df kernelmanagerd[112:abb5] Picked matching dext for bundle identifier com.company.Dext-Loader.dext: Dext com.company.Dext-Loader.dext v34 in executable dext bundle com.company.Dext-Loader.dext at /Library/SystemExtensions/B1BF8CDC-CB24-4F25-A8CA-D7A60D814861/com.company.Dext-Loader.dext.dext 2024-07-30 15:54:02.389 I kernel[0:ae71] igmp_domifreattach: reattached igmp_ifinfo for ifp XHC 2024-07-30 15:54:02.389 I kernel[0:ae71] mld_domifreattach: reattached mld_ifinfo for ifp XHC2 2024-07-30 15:54:02.389 Df kernelmanagerd[112:abb5] DextRecordTable read from plist: { com.company.Dext-Loader.dext: MRS-> Optional(( path: /Library/SystemExtensions/B1BF8CDC-CB24-4F25-A8CA-D7A60D814861/com.company.Dext-Loader.dext.dext; state: loaded )) history-> [ ( path: /Library/SystemExtensions/B1BF8CDC-CB24-4F25-A8CA-D7A60D814861/com.company.Dext-Loader.dext.dext; state: loaded ) ] } 2024-07-30 15:54:02.389 Df kernelmanagerd[112:abb5] Launching dext com.company.Dext-Loader.dext com.company.Dext-Loader.dext 0x100001464 a0cf49ca3ea45f5d54a3e8644e2dde6b0e8666c649c1e9513ca4166919038b53 2024-07-30 15:54:02.390 I kernelmanagerd[112:abb5] [com.apple.km:DextLaunch] Skipping addBreadcrumbForDextWithIdentifier for <private> 0 2024-07-30 15:54:02.389 Df kernel[0:ae71] ifnet_attach: Waiting for all kernel threads created for interface XHC2 to get scheduled at least once. 2024-07-30 15:54:02.389 Df kernel[0:ae71] ifnet_attach: All kernel threads created for interface XHC2 have been scheduled at least once. Proceeding. 2024-07-30 15:54:02.390 Df kernelmanagerd[112:abb5] Launching driver extension: Dext com.company.Dext-Loader.dext v34 in executable dext bundle com.company.Dext-Loader.dext at /Library/SystemExtensions/B1BF8CDC-CB24-4F25-A8CA-D7A60D814861/com.company.Dext-Loader.dext.dext 2024-07-30 15:54:02.479 E kernel[0:a9fb] (Sandbox) 1 duplicate report for Sandbox: imagent(633) deny(1) mach-lookup com.apple.contactsd.persistence 2024-07-30 15:54:02.479 E kernel[0:a9fb] (Sandbox) Sandbox: taskgated-helper(2985) deny(1) user-preference-read kCFPreferencesAnyApplication 2024-07-30 15:54:02.483 Df kernel[0:ae73] (AppleMobileFileIntegrity) AMFI: code signature validation failed. 2024-07-30 15:54:02.483 Df kernel[0:ae73] (AppleMobileFileIntegrity) AMFI: bailing out because of restricted entitlements. 2024-07-30 15:54:02.483 Df kernel[0:ae73] (AppleMobileFileIntegrity) AMFI: When validating /Library/SystemExtensions/B1BF8CDC-CB24-4F25-A8CA-D7A60D814861/com.company.Dext-Loader.dext.dext/com.company.Dext-Loader.dext: Code has restricted entitlements, but the validation of its code signature failed. Unsatisfied Entitlements: 2024-07-30 15:54:02.483 Df kernel[0:ae73] mac_vnode_check_signature: /Library/SystemExtensions/B1BF8CDC-CB24-4F25-A8CA-D7A60D814861/com.company.Dext-Loader.dext.dext/com.company.Dext-Loader.dext: code signature validation failed fatally: When validating /Library/SystemExtensions/B1BF8CDC-CB24-4F25-A8CA-D7A60D814861/com.company.Dext-Loader.dext.dext/com.company.Dext-Loader.dext: Code has restricted entitlements, but the validation of its code signature failed. Unsatisfied Entitlements: 2024-07-30 15:54:02.483 Df kernel[0:ae73] validation of code signature failed through MACF policy: 1 2024-07-30 15:54:02.483 Df kernel[0:ae73] check_signature[pid: 2984]: error = 1 2024-07-30 15:54:02.483 Df kernel[0:ae73] proc 2984: load code signature error 4 for file "com.company.Dext-Loader.dext" 2024-07-30 15:54:02.485 Df kernelmanagerd[112:abb5] [com.apple.libxpc.OSLaunchdJob:all] <OSLaunchdJob | handle=46B92B57-A90A-4EBD-8EF4-54313C6EE332>: submitAndStart completed, info=spawn failed, error=162: Codesigning issue 2024-07-30 15:54:02.483 Df kernel[0:ae73] (Sandbox) /Library/SystemExtensions/B1BF8CDC-CB24-4F25-A8CA-D7A60D814861/com.company.Dext-Loader.dext.dext/com.company.Dext-Loader.dext[2984] ==> com.apple.dext 2024-07-30 15:54:02.485 E kernelmanagerd[112:abb5] [com.apple.libxpc.OSLaunchdJob:all] <OSLaunchdJob | handle=46B92B57-A90A-4EBD-8EF4-54313C6EE332>: job failed to spawn, plist={ ProcessType => Driver _ManagedBy => com.apple.kernelmanagerd CFBundleIdentifier => com.company.Dext-Loader.dext _JetsamPropertiesIdentifier => com.company.Dext-Loader.dext LimitLoadToSessionType => System _DextCheckInPort => <mach send right: 0xbd486ccc0> { name = 15679, right = send, urefs = 2 } UserName => _driverkit _NullBootstrapPort => true ReslideSharedCache => false LaunchOnlyOnce => true Label => com.company.Dext-Loader.dext-0x100001464 RunAtLoad => true ProgramArguments => [<capacity = 8> 0: /Library/SystemExtensions/B1BF8CDC-CB24-4F25-A8CA-D7A60D814861/com.company.Dext-Loader.dext.dext/com.company.Dext-Loader.dext 1: com.company.Dext-Loader.dext 2: 0x100001464 3: com.company.Dext-Loader.dext ] SandboxProfile => com.apple.dext } The Xcode project uses these signing options: Automatically manage signing Team: Company Provisioning Profile: Xcode Managed Profile Signing Certificate: Apple Development: () The same project, with the same signing options, builds and loads its dext without issues from Xcode 15.3 on Sonoma 14.5. That same dext binary from Xcode 15.3 loads and passes the signature checks on Sequoia, but using Xcode on Sequoia is when the signature validation fails. Can anyone suggest a way to resolve these signature validation errors? (Other than just developing on Sonoma and testing on Sequoia?)
0
0
295
Aug ’24
Apple store connect requests sandbox entitlement for the PCI DriverKit System Extension
Hi, I have a PCI DriverKit System Extension project that our team has tested, and the entitlements are not a problem. Once we decided to place the project to the Apple Store the review team requested to add "App Sandbox" entitlement to the project. Then I added the entitlement manually to the ".entitlements" file ( I couldn't do that using the Xcode add entitlement section because since it is a driverkit project, the "App sandbox" is not visible in the entitlements page ) and re-packaged the project for distribution. Later on, I saw that the entitlement was removed during the packaging process. I also tried to add that using the "build settings" page in Xcode (the signing section ), but I had no luck. I feel like I'm being misled by the review team. Do you know if the "App sandbox" entitlement is applicable to a DriverKit project ?
3
1
643
Jul ’24
on-line entitlement format to obtain relevant "transport.usb|idVendor"
The situation: 1A) On last week our development team applied on-line for three idVendors with the format as the following: USB-VendorID: 547h(1351), 3488h(13448),4B4h(1204) Describe your apps: it shall be functional equivalence of existing Android-SDK for X-ray sensor-models:...with 15..25 text lines of the description below 1B) We received auto-reply, that we granted all the the entitlements. We've expected to have entitlements like approx like the these: &lt;key&gt;com.apple.developer.driverkit.transport.usb&lt;/key&gt; &lt;array&gt; &lt;dict&gt; &lt;key&gt;idVendor&lt;/key&gt; &lt;integer&gt;1351&lt;/integer&gt; &lt;/dict&gt; &lt;dict&gt; &lt;key&gt;idVendor&lt;/key&gt; &lt;integer&gt;131448&lt;/integer&gt; &lt;/dict&gt; &lt;dict&gt; &lt;key&gt;idVendor&lt;/key&gt; &lt;integer&gt;1204&lt;/integer&gt; &lt;/dict&gt; &lt;/array&gt; 1C) Unexpectingly (for us) ASCII-strings-fragment from our iOS-target-binary "com.rayence.dexta.driver" (built with automatic signing; seemingly this shall be our actual granted prvisionintg profile) looks as the next: &lt;plist version="1.0"&gt; &lt;dict&gt; &lt;key&gt;com.apple.developer.driverkit&lt;/key&gt; &lt;true/&gt;&lt;key&gt;com.apple.developer.driverkit.transport.usb&lt;/key&gt; &lt;array&gt;&lt;dict&gt; &lt;key&gt;idVendor&lt;/key&gt; &lt;string&gt;*&lt;/string&gt; &lt;/dict&gt;&lt;/array&gt; &lt;/dict&gt; 1D) Repectively, now we shall specify (in our dext's entitlements in XCOde-project) &lt;key&gt;com.apple.developer.driverkit.transport.usb&lt;/key&gt; &lt;array&gt; &lt;dict&gt; &lt;key&gt;idVendor&lt;/key&gt; &lt;string&gt;*&lt;/string&gt; &lt;/dict&gt; &lt;/array&gt; In DExt's info.plist we have "normal": &lt;key&gt;IOKitPersonalities&lt;/key&gt; &lt;dict&gt; &lt;key&gt;IOKitPersonality_EzHD&lt;/key&gt; &lt;dict&gt; blah-blah &lt;key&gt;idVendor&lt;/key&gt;&lt;integer&gt;1351&lt;/integer&gt; &lt;key&gt;IOKitPersonality_EzR&lt;/key&gt; &lt;dict&gt; blah-blah &lt;key&gt;idVendor&lt;/key&gt;&lt;integer&gt;131448&lt;/integer&gt; &lt;key&gt;IOKitPersonality_EzR&lt;/key&gt; &lt;dict&gt; blah-blah &lt;key&gt;idVendor&lt;/key&gt;&lt;integer&gt;1204&lt;/integer&gt; blah-blah Our ambiguities: 2a) Without exact "1D)", our DExt-installer-application never starts in iPAD. OS-loader generates obectievly-meaningless error-messages and seemingly-meaninglessly suggest to look "system-logs" (which seemingly don't contain relevant information) 2b) With exatct "1D)", 2ba) the reaction (presence/absence and timing of relevant console.app's messages) of our iPAD 17.4 on plugging/unplugging our gadgets look "non-logical". It isn't completly wrong but suspicious. E.g. our Swift-module randomly receive/dontreceive relevant events and can't connect communicate driver's instance for newly-attached physical gadgets. We realize, thses it could be our bugs; but we need disambiguation to debug the issues (or at least to be sure that the quirks are like in production system). 2bd) we would prefer debug/develop rather with real "idVendor=number nnn"s in provisioning profile (than with "idVendor=string *"). So we could be sure that quircs "2ba" are like in a production system. Respective question: Which format in on-line application ~1A) shall we use to obtain provisioning profile 1B)
2
0
508
Jul ’24
Kernel Panic triggered by PCIe DriverKit Extension
My company has been developing a DriverKit Extension for our ThunderBolt attached devices. Our testing has gone well on Intel based machines but it seems to be able to cause kernel panics on my M1/M2 Mac when click "Allow" in Privacy & Security. It always crash immediately. Unfortunately it is " ApplePPMCPMS: Could not register client id 5. Error code 0xe00002bc " and I have no way to debug it. Has anyone deal with something like this? panic(cpu 5 caller 0xfffffe002fd92b7c): "ApplePPM: virtual IOReturn ApplePPMCPMS::callPlatformFunction(const OSSymbol *, bool, void *, void *, void *, void *):2003 " "ApplePPMCPMS: Could not register client id 5. Error code 0xe00002bc\n" @ApplePPMCPMS.cpp:2003 Debugger message: panic Memory ID: 0x6 OS release type: User OS version: 22G90 Kernel version: Darwin Kernel Version 22.6.0: Wed Jul 5 22:22:52 PDT 2023; root:xnu-8796.141.3~6/RELEASE_ARM64_T8103 Fileset Kernelcache UUID: 593DE91B6B9F1B49F8178C99AC56A8C2 Kernel UUID: CE831117-201E-35F6-A293-FCC0F02097A3 Boot session UUID: EB9DE635-C521-4672-9251-955CEC7BC487 iBoot version: iBoot-8422.141.2 secure boot?: YES roots installed: 0 Paniclog version: 14 KernelCache slide: 0x00000000264e4000 KernelCache base: 0xfffffe002d4e8000 Kernel slide: 0x00000000264ec000 Kernel text base: 0xfffffe002d4f0000 Kernel text exec slide: 0x00000000278a4000 Kernel text exec base: 0xfffffe002e8a8000 mach_absolute_time: 0x3b74899144 Epoch Time: sec usec Boot : 0x66176032 0x000c0192 Sleep : 0x00000000 0x00000000 Wake : 0x00000000 0x00000000 Calendar: 0x661789be 0x0006abb4 Zone info: Zone map: 0xfffffe10002b4000 - 0xfffffe30002b4000 . VM : 0xfffffe10002b4000 - 0xfffffe14ccf80000 . RO : 0xfffffe14ccf80000 - 0xfffffe1666918000 . GEN0 : 0xfffffe1666918000 - 0xfffffe1b335e4000 . GEN1 : 0xfffffe1b335e4000 - 0xfffffe20002b0000 . GEN2 : 0xfffffe20002b0000 - 0xfffffe24ccf7c000 . GEN3 : 0xfffffe24ccf7c000 - 0xfffffe2999c48000 . DATA : 0xfffffe2999c48000 - 0xfffffe30002b4000 Metadata: 0xfffffe3efce64000 - 0xfffffe3f04e64000 Bitmaps : 0xfffffe3f04e64000 - 0xfffffe3f06478000 Extra : 0 - 0 CORE 0 recently retired instr at 0xfffffe002ea385fc CORE 1 recently retired instr at 0xfffffe002ea385fc CORE 2 recently retired instr at 0xfffffe002ea385fc CORE 3 recently retired instr at 0xfffffe002ea385fc CORE 4 recently retired instr at 0xfffffe002ea385fc CORE 5 recently retired instr at 0xfffffe002ea36edc CORE 6 recently retired instr at 0xfffffe002ea385fc CORE 7 recently retired instr at 0xfffffe002ea385fc TPIDRx_ELy = {1: 0xfffffe166798e020 0: 0x0000000000000005 0ro: 0x0000000000000000 } CORE 0 PVH locks held: None CORE 1 PVH locks held: None CORE 2 PVH locks held: None CORE 3 PVH locks held: None CORE 4 PVH locks held: None CORE 5 PVH locks held: None CORE 6 PVH locks held: None CORE 7 PVH locks held: None CORE 0: PC=0xfffffe002e919618, LR=0xfffffe002e90f630, FP=0xfffffe5d6457bc00 CORE 1: PC=0xfffffe002e936000, LR=0xfffffe002e936000, FP=0xfffffe5d6429bef0 CORE 2: PC=0xfffffe002eaacaa8, LR=0xfffffe002eaacae8, FP=0xfffffe5d63fe3690 CORE 3: PC=0xfffffe002e936000, LR=0xfffffe002e936000, FP=0xfffffe5d64693ef0 CORE 4: PC=0xfffffe002e92fbac, LR=0xfffffe002e914b70, FP=0xfffffe5d65befc60 CORE 5 is the one that panicked. Check the full backtrace for details. CORE 6: PC=0x0000000199b5b5b4, LR=0x000000018c06db24, FP=0x000000016f324ff0 CORE 7: PC=0xfffffe002ef4937c, LR=0xfffffe002ef492f0, FP=0xfffffe5d657e7de0 Compressor Info: 0% of compressed pages limit (OK) and 0% of segments limit (OK) with 0 swapfiles and OK swap space Panicked task 0xfffffe24cd09a738: 0 pages, 970 threads: pid 0: kernel_task Panicked thread: 0xfffffe166798e020, backtrace: 0xfffffe5d643735d0, tid: 62085 lr: 0xfffffe002e8fe5fc fp: 0xfffffe5d64373650 lr: 0xfffffe002ea3e6e4 fp: 0xfffffe5d64373670 lr: 0xfffffe002ea2fe4c fp: 0xfffffe5d643736e0 lr: 0xfffffe002ea2e518 fp: 0xfffffe5d643737a0 lr: 0xfffffe002e8af7f8 fp: 0xfffffe5d643737b0 lr: 0xfffffe002e8fdee4 fp: 0xfffffe5d64373b60 lr: 0xfffffe002f07a548 fp: 0xfffffe5d64373b80 lr: 0xfffffe002fd92b7c fp: 0xfffffe5d64373d30 lr: 0xfffffe002f13c88c fp: 0xfffffe5d64373d80 lr: 0xfffffe002ef44b08 fp: 0xfffffe5d64373de0 lr: 0xfffffe002ef4c3ac fp: 0xfffffe5d64373e50 lr: 0xfffffe002ef51cc8 fp: 0xfffffe5d64373f20 lr: 0xfffffe002e8b8e98 fp: 0x0000000000000000 Kernel Extensions in backtrace: com.apple.AGXG13G(227.7.14)[9A593B36-8560-3700-B806-A6531D4F72EC]@0xfffffe002f0be6a0->0xfffffe002f15dbbf dependency: com.apple.driver.AppleARMPlatform(1.0.2)[5478478E-CF49-3A40-9437-4298C08DC081]@0xfffffe002f2123d0->0xfffffe002f263173 dependency: com.apple.driver.AppleMobileFileIntegrity(1.0.5)[28D5832D-79F4-3C81-AA70-0775A365BA35]@0xfffffe002fca3580->0xfffffe002fccf587 dependency: com.apple.driver.RTBuddy(1.0.0)[8695C672-1FB4-3954-A525-C10AE1CE4E57]@0xfffffe0031348720->0xfffffe003138444b dependency: com.apple.iokit.CoreAnalyticsFamily(1)[38EC9902-05F4-3412-8CB3-1B5001455BCA]@0xfffffe00303385f0->0xfffffe0030340407 dependency: com.apple.iokit.IOGPUFamily(65.60.1)[683A153B-7683-363B-B6E8-6DC60D57D55F]@0xfffffe0030a85180->0xfffffe0030abab17 dependency: com.apple.iokit.IOReportFamily(47)[FE904E47-8D35-3152-8035-2BFA7B69A77E]@0xfffffe0030d44820->0xfffffe0030d478bf dependency: com.apple.iokit.IOSurface(336.60.1)[64104E9B-CF29-3BAC-ABF7-134DCEE9A195]@0xfffffe0030e15d20->0xfffffe0030e450e3 dependency: com.apple.kec.Libm(1)[BA66FD46-3E12-378D-A1A3-67C765DB2A21]@0xfffffe00313313c0->0xfffffe0031334e4b com.apple.driver.ApplePassthroughPPM(3.0)[495FD94A-224A-3C8B-B377-925FF62F0257]@0xfffffe002fd7ac40->0xfffffe002fdbb7c3 dependency: com.apple.driver.AppleARMPlatform(1.0.2)[5478478E-CF49-3A40-9437-4298C08DC081]@0xfffffe002f2123d0->0xfffffe002f263173 dependency: com.apple.driver.ApplePMGR(1)[8202FFB5-EBDB-333D-9001-9A2582407585]@0xfffffe002fd26380->0xfffffe002fd6c30f dependency: com.apple.iokit.IOReportFamily(47)[FE904E47-8D35-3152-8035-2BFA7B69A77E]@0xfffffe0030d44820->0xfffffe0030d478bf dependency: com.apple.kec.Libm(1)[BA66FD46-3E12-378D-A1A3-67C765DB2A21]@0xfffffe00313313c0->0xfffffe0031334e4b last started kext at 254801217669: com.apple.filesystems.exfat 1.4 (addr 0xfffffe002de1c690, size 6208) last stopped kext at 252590757699: com.apple.filesystems.exfat 1.4 (addr 0xfffffe002de1c690, size 6208)
0
0
573
Apr ’24
How to properly convert any size of application memory into the kernel space of Driverkit?
Hardware and software configuration MacBook Air M2 2022 16GB, MacOS Ventura 13.2.1 Full description This is a DriverKit that controls PCIE FPGA devices for low-latency data exchange. This driver has been implemented on Iokit, and now it needs to be launched on Driverkit to adapt to newer Macs. Driverkit lacks the IOMemoryDescriptor::withAddressRange(Iokit) function to convert the app's memory of any size to a Descriptor. Currently, we use args->structureOutputDescriptor->CreateMapping to map the Descriptor passed by the application to the kernel layer. // App size_t ***::xxRead(long long addr, size_t size, void * buff){ std::lock_guard<std::mutex> guard(usrLock); kern_return_t kr; uint64_t info[2] = {(uint64_t)addr, (uint64_t)size}; kr = IOConnectCallMethod( connect, kUserReadIO, info, 2, NULL, NULL, NULL, NULL, buff, &size); return size; } // Driverkit const IOUserClientMethodDispatch sMethods[kNumMethods] = { [kUserReadIO] = { (IOUserClientMethodFunction) &SmiPcieUc::sUserReadIo, .checkCompletionExists = false, .checkScalarInputCount = 2, // Read Addr, size .checkStructureInputSize = 0, .checkScalarOutputCount = 0, .checkStructureOutputSize = kIOUserClientVariableStructureSize} // Read Data }; kern_return_t SmiPcieUc::sUserReadIo (OSObject * target, void* reference, IOUserClientMethodArguments* args){ IOMemoryMap * memMap = nullptr; uint32_t * buffKptr = nullptr; kern_return_t rt = 0; if(target == nullptr){ Log("***Err***: sUserReadIo Target is Null!"); return kIOReturnError; } if(args->structureOutputDescriptor){ rt = args->structureOutputDescriptor->CreateMapping(0,0,0,0,0, &memMap); if(rt == kIOReturnSuccess){ buffKptr = reinterpret_cast<uint32_t *>(memMap->GetAddress()); } else { Log("***Err***: sUserReadIo Mapping Failed!"); return kIOReturnNoMemory; } } else { buffKptr = (uint32_t *) args->structureOutput; } rt = ((SmiPcieUc *)target)->UserReadIo((uint64_t *)&args->scalarInput[0], (size_t *)&args->scalarInput[1], buffKptr); OSSafeReleaseNULL(memMap); return rt; } phenomenon When StructureOutputSize is greater than 4096, args>structureOutputDescriptor exists, and when it is less than or equal to 4096, args->structureOutputDescriptor and args->structureOutput are both equal to nullptr, (in IOkit, args->structureOutput is not empty)。 How to properly convert any size of application memory into the kernel space of Driverkit?
1
0
618
Mar ’24
How to get access of NVMe Controller Admin submission and completion queues?
We have been doing a R&D work related to the NVMe controller on Mac platform, where we need to get control of the admin queues(submission as well as completion). From the spec of NVMe it’s very clear that what are all registers do we need to deal with to get access of the queues. We are accordingly following those registers to create our own queues. Also we have prepared and enqueued a sample admin command to the newly created submission queue. But surprisingly we can’t get any assurance whether the command got processed by the controller or not, because from the completion queue entry we can see all the entries are zero, which is not expected anyway. So here the question is, how to communicate with the controller properly ? We are also aware of the fact of existing NVMe driver(IONVMeFamily) on Mac platform, is this somehow crossing our way ? We have done all the proper setup for registers, DMAs and interrupt. Path is very ok if we use builtin driver with the XNVME user space application (we can trigger limited admin commands over there). But here we need to have our own created queues up and running with seamless admin command transaction. Here we must tell about our setup, we have one SSD which is connected via a thunderbolt cable to Mac laptop using type C usb port. We have tried to access pre-configured admin queues from IONVMeFamily driver but that is also a blocker for us, as we can’t see any valid data from submission/completion queues. Request you to all please help us coming out of this trapped zone.
0
0
499
Mar ’24
Write Dext code for my existing Kext
Hi,I am trying to write Dext code for my existing Kext,How to convert this code to be compatible with Dext? BufferMemoryDescriptorAME_Module = NULL; IOMemoryMap *MemMap; BufferMemoryDescriptorAME_Module= IOBufferMemoryDescriptor::inTaskWithPhysicalMask(kernel_task,kIOMemoryPhysicallyContiguous,otal_memory_size); BufferMemoryDescriptorAME_Module->prepare(kIODirectionInOut); MemMap = BufferMemoryDescriptorAME_Module->map(kIOMapInhibitCache); logicalAddressAME_Module = (UInt8 *) MemMap->getVirtualAddress(); physicalAddressAME_Module = MemMap->getPhysicalAddress(); Thanks, Frederic
0
0
492
Feb ’24
Shared data queue between dext and app (IOSharedDataQueueDispatchSource?)
I'm looking for a fast and efficient way for user-space to send I/O to my driver. One way I'd have hoped to do this, was through a shared memory ring-buffer. In the WWDC19 presentation on System Extensions and DriverKit, at roughly 17:00, they mention an IOSharedDataQueueDispatchSource. This doesn't exist in the DriverKit API. An IODataQueueDispatchSource is available, but doesn't seem to be meant to be shared. In the old IOKit framework, there are similar IOSharedDataQueue and IODataQueue, but they are unavailable in DriverKit. So, what are my options for implementing a fast, efficient I/O path to my driver?
1
1
652
Jan ’24
Missing API function in DriverKit for canceling timer event source
Hi, I'm working on a project that requires porting my KEXT driver to DriverKit system extension ( DEXT ). I've managed to re-write most of functions using DriverKit but I couldn't found a function for cancelling a scheduled timer. What I try to make is waiting for an interrupt with a timeout period. And if the interrupt arrives in certain time period I need to cancel timeout timer and send async response to the user app. I'm also open to the suggestions. Here is the code fragment for IOKit that I need to convert to DriverKit // ... initialization code fragment ( error handling parts deleted ) m_timeout_timer = IOTimerEventSource::timerEventSource(this, timer_fired_pc_recv); IOReturn result = m_workLoop->addEventSource(timer); // ... starting the timeout counter m_timeout_timer->setTimeoutMS(timeout_in_ms); // ... some time later (cancel part) m_timeout_timer->cancelTimeout(); And equivalent code in DriverKit // ... initialization code fragment ( error handling parts deleted ) ret = IOTimerDispatchSource::Create(ivars->default_dispatch_queue, &ivars->timer_dispatchSource); // starting timer timer_dispatchSource->WakeAtTime(kIOTimerClockMonotonicRaw, currentTime + fiveSecondsInNanoSeconds, twoSecondsInNanoSeconds); // missing part here I did see the member method Cancel in IOTimerDispatchSource but it is kind of termination to the lifetime of IOTimerDispatchSource
0
0
511
Oct ’23
Unsatisfied entitlements: com.apple.developer.driverkit.transport.pci
Hi, I'm developing my own PCIe device driver, the log shows error message below when driver executing. 2023-07-31 13:43:47.031012+0800 0x1d41ce Error 0x0 12158 0 taskgated-helper: (ConfigurationProfiles) [com.apple.ManagedClient:ProvisioningProfiles] com.asix.dext.pciedevice: Unsatisfied entitlements: com.apple.developer.driverkit.transport.pci 2023-07-31 13:43:47.031048+0800 0x1d41ce Error 0x0 12158 0 taskgated-helper: (ConfigurationProfiles) [com.apple.ManagedClient:ProvisioningProfiles] Disallowing: com.asix.dext.pciedevice 2023-07-31 13:43:47.062775+0800 0x1d436e Error 0x0 103 0 kernelmanagerd: Error occurred while handling request "DextLaunch(arguments: Optional(["Check In Token": 12087, "kOSBundleDextUniqueIdentifier": <04642bc8 90788071 c2a02259 c624ba81 3bebbf55 f9f2db7e f9fbbdd5 1f2ed99d>, "Driver Extension Server Tag": 4294982732, "DriverKit Reslide Shared Cache": 0, "Driver Extension Server Name": com.asix.dext.pciedevice, "CFBundleIdentifier": com.asix.dext.pciedevice]))": Error Domain=NSPOSIXErrorDomain Code=8 "Exec format error" **How to match device's PID & VID to driver and make it work successfully? Please help me, Thanks** errorcode.txt
2
0
1.1k
Nov ’23
PCI dext: scatter-gather DMA to application buffer
Hello, I am porting a PCI driver written with IOKit to the new PCIDriverKit framework. I am able to perform DMA with a contiguous buffer allocated inside the dext (with IOBufferMemoryDescriptor::Create). But I would also like to perform DMA to and from a buffer allocated by an application. What I exactly want to do is: Allocate an aligned buffer in the app (with e.g. posix_memalign) Send the pointer to this buffer to the dext In the dext, retrieve the list of pages descriptors to be able to perform DMA without copy into (or from) this buffer. In IOKit, we can use methods such IOMemoryDescriptor::withAddressRange and IODMACommand::gen64IOVMSegments to map and retrieve the scatter gather list but I cannot find any information on how to do this in a dext with the PCIDriverKit framework. Can anybody help me on how to do that?
4
0
2.2k
Oct ’23