users

users@lists.genode.org

2285 discussions

interrupt issues caused by multiple IO-APICs?
by Nobody III 12 Sep '18

12 Sep '18

I've been trying to debug Genode's interrupt issues on my desktop computer on and off for a few years now, and I've recently gathered more information by testing some different devices and configurations that I hadn't before. I'd already found that both usb_drv and ahci_drv waited for interrupts that never came, and were thus unable to function properly. However, fb_drv works without any issues. I know from previous debugging, with the help of multiple Genode developers, that my computer has two IO-APICs (rather than one) and an unusual ACPI table. I've tried using ACPICA in place of Genode's custom ACPI driver, but that doesn't solve my problems, even now that it calls the _PIC method to configure the interrupt controller to run in APIC mode. Recently, I have acquired a PS/2 mouse and keyboard, and have found, to my surprise, that they worked properly in Genode, unlike both USB and AHCI. I was even more surprised to find that my recently-added Intel 8260 wireless card also worked under Genode. I examined my interrupt table (available at https://gist.github.com/NobodyIII/573e58bcb82446ed1da36043676c3013), and found that the wireless card uses MSI interrupts, and noticed that the XHCI controller also uses MSI interrupts. I modified Sculpt's usb_drv config to disable everything except XHCI, and plugged my USB mouse and keyboard into USB 3.0 ports, and that allowed me to use them as well. After further examination of my interrupt table, I noticed that the i8042-compatible PS/2 controller uses regular IO-APIC interrupts like the ACHI, OHCI, and EHCI controllers do. However, AHCI, OCHI, and EHCI all use the upper half of my IRQs (16-32), while the PS/2 controller only uses the lower half. Assuming that the two IO-APICs split the IRQs such that one handles the lower half of the IRQs and the other handles the upper half, it seems that Genode is likely handling the first IO-APIC correctly, but not the second one. Does anyone have any more information or ideas? And has anyone else tried Genode on a computer with multiple IO-APICs?

1 0

rump_cgd
by Steven Harp 11 Sep '18

11 Sep '18

Is rump_cgd (block encrypted devices introduced ~14.05) still supported/operational in Genode 18.x? Issue #2224 mentions "not converted" in transition to new base API [1], and #2778 seems to have removed the 'tool/rump' script (which was handy to prepare encrypted volumes) [2]. // Steve Harp [1] https://github.com/genodelabs/genode/issues/2224 [2] https://github.com/genodelabs/genode/issues/2778

3 3

Boot problem on tz_vmm demo for imx53-qsb
by lzSun 11 Sep '18

11 Sep '18

Hello, I am using genode 18.05 and I am trying the TZ_VMM demo on imx53-qsb board. I encountered problem when the system is booting. The error like this: [init -> tz_vmm] Start virtual machine ... [init -> tz_vmm] Error: failed to handle data abort [init -> tz_vmm] Cpu state: [init -> tz_vmm] Register Virt Phys [init -> tz_vmm] ------------------------------------ [init -> tz_vmm] r0 = 0x07000000 0x00000000 [init -> tz_vmm] r1 = 0xf801ff50 0x00000000 [init -> tz_vmm] r2 = 0xffffffe4 0x00000000 [init -> tz_vmm] r3 = 0x40002000 0x00000000 [init -> tz_vmm] r4 = 0x00000000 0x00000000 [init -> tz_vmm] r5 = 0x00000000 0x00000000 [init -> tz_vmm] r6 = 0x000000ee 0x00000000 [init -> tz_vmm] r7 = 0x00000000 0x00000000 [init -> tz_vmm] r8 = 0x00000000 0x00000000 [init -> tz_vmm] r9 = 0x00000000 0x00000000 [init -> tz_vmm] r10 = 0x00000000 0x00000000 [init -> tz_vmm] r11 = 0x00000000 0x00000000 [init -> tz_vmm] r12 = 0x00000000 0x00000000 [init -> tz_vmm] sp = 0x00000000 0x00000000 [init -> tz_vmm] lr = 0x00000000 0x00000000 [init -> tz_vmm] ip = 0x0040dfe8 0x00000000 [init -> tz_vmm] cpsr = 0x60000093 0x00000000 [init -> tz_vmm] sp _und = 0x00000000 0x00000000 [init -> tz_vmm] lr _und = 0x80008008 0x00000000 [init -> tz_vmm] spsr _und = 0x00000093 0x00000000 [init -> tz_vmm] sp _svc = 0xf801ff4c 0x00000000 [init -> tz_vmm] lr _svc = 0x0040bc97 0x00000000 [init -> tz_vmm] spsr _svc = 0x00000000 0x00000000 [init -> tz_vmm] sp _abt = 0x00000000 0x00000000 [init -> tz_vmm] lr _abt = 0x00000000 0x00000000 [init -> tz_vmm] spsr _abt = 0x00000000 0x00000000 [init -> tz_vmm] sp _irq = 0x00000000 0x00000000 [init -> tz_vmm] lr _irq = 0x00000000 0x00000000 [init -> tz_vmm] spsr _irq = 0x00000000 0x00000000 [init -> tz_vmm] sp _fiq = 0x00000000 0x00000000 [init -> tz_vmm] lr _fiq = 0x00000000 0x00000000 [init -> tz_vmm] spsr _fiq = 0x00000000 0x00000000 [init -> tz_vmm] ttbr0 = 0x00000000 [init -> tz_vmm] ttbr1 = 0x00000000 [init -> tz_vmm] ttbrc = 0x00000000 [init -> tz_vmm] dfar = 0x07000000 0x00000000 [init -> tz_vmm] exception = data_abort And my steps are: 1. get linux image: git clone https://github.com/skalk/linux.git git checkout imx53-tz cross_compile and get uImage 2.compile genode: ./tool/create_builddir hw_imx53_qsb_tz add RUN_OPT = --include boot_dir/hw --include image/uboot copy uImage to bin/linux comment out "exec >& /dev/null wget -O bin/linux.md5 $linux_uri.md5" comment out "exec md5sum -c linux.md5" make run/tz_vmm and get uImage 3.get u-boot git clone https://github.com/skalk/u-boot/tree/imx53_loco cross compile and get u-boot.imx 4.Booting by ext2load and bootm command. Can anyone help me fix this problem? ------------------ Best wishes

2 1

TZ_VMM demo on i.mx53
by lzSun 11 Sep '18

11 Sep '18

Hello, I am new to Genode and the version I am using is 18.05. I am trying to boot genode on the secure world and linux on normal world on i.MX53-QSB board. I follow steps in this link: https://sourceforge.net/p/genode/mailman/message/33588666/. The system can boot successfully ,but after the linux started，it periodically print this message: mmc0: Timeout waiting for hardware interrupt. mxsdhci: ============== REGISTER DUMP ============== mxsdhci: Sys addr: 0x00000000 | Version: 0x00001201 mxsdhci: Blk size: 0x00000000 | Blk cnt: 0x00000001 mxsdhci: Argument: 0x00000c00 | Trn mode: 0x341a0000 mxsdhci: Present: 0xff880008 | Host ctl: 0x00000121 mxsdhci: Clock: 0x000010ff mxsdhci: Int stat: 0x00010001 mxsdhci: Int enab: 0x107f010b | Sig enab: 0x107f000b mxsdhci: Caps: 0x07f30000 mxsdhci: =========================================== I am not sure whether this is normal, or it's because something wrong.And I have a few question about the TZ_VMM demo on i.MX53 board hope to get help: 1. I looked up the code in repos/os/src/server/tz_vmm, in target.mk file, there's only empty.cc was declared to be compiled. So how can other files can be compiled and used? 2. I want to implement a secure app in Genode OS and provide service to normal world.But I have difficult to achieve this goal. To my understanding, I should implement the server module that can be used by VMM. Also, I should design help module in the linux kernel which uses SMC instruction to trap to VMM. Is it right? Actually, in tz_vmm dictionary, I can only see a block driver and serial driver which are used to handle block and serial driver smc respectively. So I don't know where can I add my code to serve normal world's smc request. Is there any suggestion or guidance that I can refer to? Best Regards,

2 1

Genode on AMD SB7x0/SB8x0/SB9x0
by Nobody III 11 Sep '18

11 Sep '18

I've had issues with interrupts on Genode on my desktop computer with an AMD chipset. In particular, the USB and AHCI drivers both get stuck waiting for interrupts. However, the VESA framebuffer driver works fine. I have tried using ACPICA to solve my issues, but with no success. I recently bought a PS/2 mouse, and it works just fine. Also, I recently installed an Intel wireless card, which Sculpt was able to detect. When I selected WiFi, the driver was able to detect nearby wireless networks. I have not been able to test whether I can connect yet, because I don't have a PS/2 keyboard. I'll try inserting my own wifi_config XML file with an SSID and password into my Sculpt image tomorrow, and then I can report on whether that works. What does this new information indicate?

1 0

openssl port
by Steven Harp 10 Sep '18

10 Sep '18

Is there some chance that the OpenSSL port will be upgraded in the near future? This port currently references * 1.0.1i *, which is no longer supported. (Versions 1.0.2p or 1.1.0i are still supported.) Given OpenSSL's relevance to features like native stub resolver, this might be a good opportunity to upgrade? // Steve Harp

1 0

Sculpt/wifi_drv regression: no IP address
by Nobody III 10 Sep '18

10 Sep '18

I updated to the 18.08 Genode release, built a Sculpt image, then booted it on my Dell Latitude 6430u laptop. I discovered that the WiFi driver, which worked in 18.05 (Sculpt TC) no longer works properly. In particular, the WiFi driver detects the access points and can connect to them, but doesn't get an IP address. I tried multiple access points, both my wireless router and my phone. With my phone, I used a clean lowercase alphabetic SSID ("bensphone"), to avoid any issues with problematic characters. I am using an Intel Ultimate-N 6300 wireless card.. Wired networking still works.

2 3

Checking WiFi driver
by Prakash A 10 Sep '18

10 Sep '18

Hi How to check whether WiFi driver is working or not. I mean what is run command. Regards, Prakash A

2 1

missing symbols for qt5_widgets
by Nobody III 07 Sep '18

07 Sep '18

When linking object code that includes a custom subclass of QPlainTextEdit, I get the following error: undefined reference to `typeinfo for QPlainTextEdit' I have been able to solve this problem by keeping more of the qt5_widgets symbols than those available in the depot API package, which are obtained from https://github.com/cproc/genode-symbols. When looking at the symbols file, I see that the typeinfo and vtable symbols were manually added for some Qt5 widget classes. Given the commonly-used ability to inherit Qt5 classes, wouldn't it be best to keep these symbols for all classes? The classes with names that end in Private could be split into a separate API package for compatibility and/or to reduce the size of the symbols lists and the resulting .lib.so files.

1 0

some recipe hashes need to be updated
by Nobody III 06 Sep '18

06 Sep '18

I noticed while building Qt5 packages for Genode 18.08 that mesa and qt5_dejavusans needed their hashes updated before I could build them. Also, I couldn't prepare the fatfs port without updating the hash for ff13b.zip and then for fatfs.port.

3 10

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

users