Hi, I''m trying to get the VTPM to work with an Atmel TPM. The computer is an IBM T41p running Debian Etch, and I verified that the TPM is working well using the libtpm library: ,---- | debian:~# ./libtpm-2.0/utils/tpm_demo | TPM successfully reset | TPM version 1.1.0.6 | 16 PCR registers are available | ... `---- I installed Xen testing from source (downloaded by `hg clone http://xenbits.xensource.com/xen-3.0-testing.hg'' yesterday), and Xen itself is running fine. I compiled my own kernel (make linux-2.6-xen-config CONFIGMODE=menuconfig, build...) with tpm_atmel and the Xen TPM options included. Before starting any DomU, i run `vtpm_managerd'', and everything looks fine: ,---- | INFO[VTPM]: Loaded saved state (dmis = 8). | INFO[VTSP]: Loading Key into TPM. | INFO[VTPM]: Re-attaching DMI instance 0 on domain 0 . | INFO[TCS]: Calling TCS_OpenContext: | INFO[VTPM]: [1]: Waiting for Guest requests & ctrl messages. | INFO[VTPM]: [2]: Waiting for DMI messages. `---- Once I start my test domain, the following messages are displayed: ,---- | INFO[VTPM]: [1]: RECV[14}: 0x0 0 0 0 1 c1 0 0 0 13 0 0 0 1 0 0 0 0 0 0 0 0 1 | INFO[VTPM]: Re-attaching DMI instance 1 on domain 0 . | INFO[TCS]: Calling TCS_OpenContext: | ERROR[VTPM]: Could not exec to launch vtpm | INFO[VTPM]: [1]: SENT: 0x0 0 0 0 1 c1 0 0 0 a 0 0 0 0 | INFO[VTPM]: [1]: Waiting for Guest requests & ctrl messages. | INFO[VTPM]: Launching DMI on PID = 7308 | ERROR[VTPM]: [1]: Error writing to BE. Aborting... | INFO[VTPM]: [1]: Waiting for Guest requests & ctrl messages. | ERROR[VTPM]: [1]: Can''t open inbound fh for /dev/vtpm. `---- (BTW: If I close vtpm_managerd with CTRL+C, vtpm_managerd segfaults and keys are left loaded in the TPM: ,---- | INFO[VTPM]: VTPM Manager shutting down for signal 2. | INFO[TCS]: Calling TCS_CloseContext. | INFO[VTPM]: Child shutting down | Segmentation fault `----) An exec of vtpm fails because there is now /usr/bin/vtpmd. However, if I modify the Makefile in tools/vtpm_manager/manager to build vtpmd (added it to the BIN= line), I get the following error: ,---- | INFO[VTPM]: [1]: RECV[14}: 0x0 0 0 0 1 c1 0 0 0 13 0 0 0 1 0 0 0 0 0 0 0 0 1 | INFO[VTPM]: Re-attaching DMI instance 1 on domain 0 . | INFO[TCS]: Calling TCS_OpenContext: | INFO[VTPM]: Starting VTPM. | INFO[TCS]: Constructing new TCS: | ERROR[TXDATA]: TPM open failedERROR in VTPM_Init_Service at vtpm_manager.c:769 code: TPM_IOERROR. | ERROR[VTPM]: Closing vtpmd due to error during startup. | INFO[VTPM]: Launching DMI on PID = 7632 | INFO[VTPM]: [1]: SENT: 0x0 0 0 0 1 c1 0 0 0 a 0 0 0 0 | INFO[VTPM]: [1]: Waiting for Guest requests & ctrl messages. `---- So it seems this is not the way it should be... If I replace /usr/bin/vtpmd with an empty shell script, I don''t get errors when starting my test domain, but on the first access (`cat /sys/bus/platform/devices/tpm_vtpm/pcrs'' or tpm_demo again): ,---- | INFO[VTPM]: [1]: RECV[14}: 0x0 0 0 0 1 c1 0 0 0 13 0 0 0 1 0 0 0 0 0 0 0 0 1 | INFO[VTPM]: Re-attaching DMI instance 1 on domain 0 . | INFO[TCS]: Calling TCS_OpenContext: | INFO[VTPM]: Launching DMI on PID = 7965 | INFO[VTPM]: [1]: SENT: 0x0 0 0 0 1 c1 0 0 0 a 0 0 0 0 | INFO[VTPM]: [1]: Waiting for Guest requests & ctrl messages. (running `cat /sys/bus/platform/devices/tpm_vtpm/pcrs'') | INFO[VTPM]: [1]: RECV[14}: 0x0 0 0 1 0 c1 0 0 0 16 0 0 0 65 0 0 0 5 0 0 0 4 0 0 1 1 | INFO[VTPM]: [1]: Forwarding command to DMI. | ERROR[VTPM]: [1]: VTPM ERROR: Can''t open outbound fh to dmi. | INFO[VTPM]: [1]: SENT: 0x0 0 0 1 0 c1 0 0 0 a 0 0 0 1f | INFO[VTPM]: [1]: Waiting for Guest requests & ctrl messages. `---- Can you tell me what I am doing wrong? Thanks in advance, Martin -- Martin Hermanowski http://martin.hermanowski.name _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users