| Message ID | 20220816142056.1222020-1-philip.cox@canonical.com |
|---|---|
| Headers | show |
| Series | i915: Enable HuC authentication for EHL, ICL, and ADL-S | expand |
On 8/16/22 08:20, Philip Cox wrote: > BugLink: https://bugs.launchpad.net/bugs/1981971 > > v2: removed magic number in patch 0002, and clean up commit message > in commit 0001. > v3: added version history to cover letter, and to patch commit messages. > > [Impact] > * In the i915 Intel gfx driver, we need to explicitly enable HuC > authentication. > > * This fixes an issue with the Gstreamer media pipeline. > > * This only applies to the ICL, ADL-S, and EHL platforms. > > * This patch is only for the linux-intel-iotg kernel. We have pulled > in patches from upstream, so the state of this driver is in between > the upstream 5.15 driver, and the upstream tip-of-tree driver. > > [Test Plan] > * Intel has tested this patch. > > * I have tested the patch as well. > > * To test, the system is booted, and issuing the command: > sudo cat /sys/kernel/debug/dri/0/gt/uc/huc_info > will either return the huc info, or it will display > "HuC disabled" > > > > [Where problems could occur] > * If the installed firmware has been trimmed so that only blobs > that were in use were retained, when the kernel tried to load > the HuC blob, it would fail. This scenario should not happen > unless someone is mucking about with the installed firmware. > If that is the case, other things will likely start failing > as well. > Acked-by: Tim Gardner <tim.gardner@canonical.com>
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA512 Acked-by: Marcelo Henrique Cerri <marcelo.cerri@canonical.com> On Tue, Aug 16 2022, Philip Cox wrote: > BugLink: https://bugs.launchpad.net/bugs/1981971 > > v2: removed magic number in patch 0002, and clean up commit message > in commit 0001. > v3: added version history to cover letter, and to patch commit messages. > > [Impact] > * In the i915 Intel gfx driver, we need to explicitly enable HuC > authentication. > > * This fixes an issue with the Gstreamer media pipeline. > > * This only applies to the ICL, ADL-S, and EHL platforms. > > * This patch is only for the linux-intel-iotg kernel. We have pulled > in patches from upstream, so the state of this driver is in between > the upstream 5.15 driver, and the upstream tip-of-tree driver. > > [Test Plan] > * Intel has tested this patch. > > * I have tested the patch as well. > > * To test, the system is booted, and issuing the command: > sudo cat /sys/kernel/debug/dri/0/gt/uc/huc_info > will either return the huc info, or it will display > "HuC disabled" > > > > [Where problems could occur] > * If the installed firmware has been trimmed so that only blobs > that were in use were retained, when the kernel tried to load > the HuC blob, it would fail. This scenario should not happen > unless someone is mucking about with the installed firmware. > If that is the case, other things will likely start failing > as well. - -- Regards, Marcelo -----BEGIN PGP SIGNATURE----- iQHQBAEBCgA6FiEExJjLjAfVL0XbfEr56e82LoessAkFAmL73M8cHG1hcmNlbG8u Y2VycmlAY2Fub25pY2FsLmNvbQAKCRDp7zYuh6ywCYR4C/4wfHcv4t6ow9CMOZMH e9mljFfZCxoVW23nfMNs/4y8eMW+TWVz1iaoiukW6lpvMf2VBrQzipJnvhlMF4in G1qGu/tLKGaaAWyf+x3nNuvvRIKUJMtvnirp+20mmcdFF+w0shgGEmf6f3CRrNAo aCfLKQ8WUlPVIweGa1QgVZucLUgLh8GHWlfK/jWE2O3rFhhE/jTQu0lsz0wF48+/ OmiJaTi1q0Vkp05ghBwGum217fYGGBo6W8ew1Rof4R0aYjJTG1mqbcEaQ7QXo/lh GfAW6OowOeKUyxD43/RIcFBxeoyCpfnWcQ8avo6UhUggQvuGkoGWXb9WtgZC21bG C53O/BBsHLRyY2paWQxPNlN/a8jb8FwoaXTNVItB8MMS3egTKCrmGCpnmw1BFjNO s5MLwI+kFEG3nJLPIHbXl/eWUV69pnXvrTo4CdL2SQPW/Gr4yFLiF98flJz91PWv /JQV6zhEvlHY3aNiYSLVAfejSSslBCKUS1JQ9C7b9XRnVj4= =saXM -----END PGP SIGNATURE-----
On Tue, 2022-08-16 at 10:20 -0400, Philip Cox wrote: > BugLink: https://bugs.launchpad.net/bugs/1981971 > > v2: removed magic number in patch 0002, and clean up commit message > in commit 0001. > v3: added version history to cover letter, and to patch commit > messages. > > [Impact] > * In the i915 Intel gfx driver, we need to explicitly enable HuC > authentication. > > * This fixes an issue with the Gstreamer media pipeline. > > * This only applies to the ICL, ADL-S, and EHL platforms. > > * This patch is only for the linux-intel-iotg kernel. We have pulled > in patches from upstream, so the state of this driver is in > between > the upstream 5.15 driver, and the upstream tip-of-tree driver. > > [Test Plan] > * Intel has tested this patch. > > * I have tested the patch as well. > > * To test, the system is booted, and issuing the command: > sudo cat /sys/kernel/debug/dri/0/gt/uc/huc_info > will either return the huc info, or it will display > "HuC disabled" > > > > [Where problems could occur] > * If the installed firmware has been trimmed so that only blobs > that were in use were retained, when the kernel tried to load > the HuC blob, it would fail. This scenario should not happen > unless someone is mucking about with the installed firmware. > If that is the case, other things will likely start failing > as well.
BugLink: https://bugs.launchpad.net/bugs/1981971 v2: removed magic number in patch 0002, and clean up commit message in commit 0001. v3: added version history to cover letter, and to patch commit messages. [Impact] * In the i915 Intel gfx driver, we need to explicitly enable HuC authentication. * This fixes an issue with the Gstreamer media pipeline. * This only applies to the ICL, ADL-S, and EHL platforms. * This patch is only for the linux-intel-iotg kernel. We have pulled in patches from upstream, so the state of this driver is in between the upstream 5.15 driver, and the upstream tip-of-tree driver. [Test Plan] * Intel has tested this patch. * I have tested the patch as well. * To test, the system is booted, and issuing the command: sudo cat /sys/kernel/debug/dri/0/gt/uc/huc_info will either return the huc info, or it will display "HuC disabled" [Where problems could occur] * If the installed firmware has been trimmed so that only blobs that were in use were retained, when the kernel tried to load the HuC blob, it would fail. This scenario should not happen unless someone is mucking about with the installed firmware. If that is the case, other things will likely start failing as well.