You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Backport newer changes to 5.4, especially the new IPTS driver. This is motivated by a discussion on Gitter. Basically, on the Pro 7 there seem to be some upstream issues with the iGPU, causing freezes on v5.5+. It would thus help some people to have patches including the latest wifi fixes and IPTS driver on v5.4, even if we don't build binary releases.
We should probably remove the old 5.4 branch and start fresh based on the 5.6 patches.
So, functionally, all of the ioremap_ucpatches are included on v5.4 (and also on v4.19 [1]) recently.
v5.4-surface-devel and v4.19-surface-devel has the additional docs patch: "docs: driver-model: add devm_ioremap_uc" but not on upstream v5.4 and v4.19 (at least yet?)
[1]
btw, the following commits are also backported into v4.19:
"mfd: intel-lpss: Use devm_ioremap_uc for MMIO" (ours: d63f165) is also backported on v4.19.122 as e06a899
"lib: devres: add a helper function for ioremap_uc" (ours: 32bfaf2) is also backported on v4.19.122 as ccc4433
Thanks! I'll probably have a go at this later. Seems like its probably easier to update the current 5.4. I think the docs commit will not be backported as its just for documentation. It's not important for functionality.
There was a merge conflict when rebasing onto v5.4.41, so I've force-pushed the rebased to 5.4-surface-devel so that we don't have to deal with that again in the future.
I built 5.4.41 Arch kernel myself. It's working properly regarding linux-surface patches.
Also, current v5.4-surface-devel branch looks good to me, except the following difference to v5.4-surface:
$ git diff v5.4-surface-devel v5.4-surface
diff --git a/Documentation/driver-api/driver-model/devres.rst b/Documentation/driver-api/driver-model/devres.rst
index 92628fdc2f11..a100bef54952 100644
--- a/Documentation/driver-api/driver-model/devres.rst+++ b/Documentation/driver-api/driver-model/devres.rst@@ -314,7 +314,6 @@ IOMAP
devm_ioport_unmap()
devm_ioremap()
devm_ioremap_nocache()
- devm_ioremap_uc()
devm_ioremap_wc()
devm_ioremap_resource() : checks resource, requests memory region, ioremaps
devm_iounmap()
diff --git a/drivers/input/touchscreen/ipts/init.c b/drivers/input/touchscreen/ipts/init.c
index fb70d55542af..6c2dcd88ce07 100644
--- a/drivers/input/touchscreen/ipts/init.c+++ b/drivers/input/touchscreen/ipts/init.c@@ -1,6 +1,5 @@
// SPDX-License-Identifier: GPL-2.0-or-later
-#include <linux/dma-mapping.h>
#include <linux/mei_cl_bus.h>
#include <linux/module.h>
#include <linux/mod_devicetable.h>
@@ -21,17 +20,6 @@ static int ipts_init_probe(struct mei_cl_device *cldev,
dev_info(&cldev->dev, "Probing IPTS\n");
- // Setup the DMA bit mask- if (!dma_coerce_mask_and_coherent(&cldev->dev, DMA_BIT_MASK(64))) {- dev_info(&cldev->dev, "IPTS using DMA_BIT_MASK(64)\n");- } else if (!dma_coerce_mask_and_coherent(&cldev->dev,- DMA_BIT_MASK(32))) {- dev_info(&cldev->dev, "IPTS using DMA_BIT_MASK(32)");- } else {- dev_err(&cldev->dev, "No suitable DMA for IPTS available\n");- return -EFAULT;- }-
ret = mei_cldev_enable(cldev);
if (ret) {
dev_err(&cldev->dev, "Cannot enable IPTS\n");
I guess docs patch for ioremap_uc is intentionally dropped?
wrt the DMA thing, those lines don't exist on v5.4-surface but exist on generated patches on linux-surface repo. So, I guess some accident happened? btw, on SB1, ipts works with those DMA lines (built kernel with patches from linux-surface repo). EDIT: ah, ok. I haven't read discussion on irc correctly. this commit linux-surface/intel-precise-touch@a3e443c caused issue on 5.4, so the commit isn't applied on 5.4. So, what is not updated is v5.4-surface.
Activity
kitakar5525 commentedon May 15, 2020
All of the commits from me that is relevant to v5.4 are applied to v5.4
I guess starting freshly may not be needed but I'll leave it to you.
Regarding the other commits, as far as I know:
backported by upstream
ioremap_uc
hid
missing
surface-sam
ipts: all of the commits are missing
kitakar5525 commentedon May 15, 2020
So, functionally, all of the
ioremap_ucpatches are included on v5.4 (and also on v4.19 [1]) recently.v5.4-surface-devel and v4.19-surface-devel has the additional docs patch: "docs: driver-model: add devm_ioremap_uc" but not on upstream v5.4 and v4.19 (at least yet?)
[1]
btw, the following commits are also backported into v4.19:
qzed commentedon May 16, 2020
Thanks! I'll probably have a go at this later. Seems like its probably easier to update the current 5.4. I think the docs commit will not be backported as its just for documentation. It's not important for functionality.
qzed commentedon May 16, 2020
v5.4-surface-develshould be up to date now. Have yet to test it.qzed commentedon May 16, 2020
There was a merge conflict when rebasing onto v5.4.41, so I've force-pushed the rebased to 5.4-surface-devel so that we don't have to deal with that again in the future.
kitakar5525 commentedon May 18, 2020
I built 5.4.41 Arch kernel myself. It's working properly regarding linux-surface patches.
Also, current
v5.4-surface-develbranch looks good to me, except the following difference tov5.4-surface:ioremap_ucis intentionally dropped?wrt the DMA thing, those lines don't exist onv5.4-surfacebut exist on generated patches on linux-surface repo. So, I guess some accident happened?btw, on SB1, ipts works with those DMA lines (built kernel with patches from linux-surface repo).EDIT: ah, ok. I haven't read discussion on irc correctly. this commit linux-surface/intel-precise-touch@a3e443c caused issue on 5.4, so the commit isn't applied on 5.4. So, what is not updated is
v5.4-surface.qzed commentedon May 18, 2020
The
ioremap_ucdoc is intentionally dropped, yes. Without the doc patch we don't need to have a separate patch forioremap_uc.I forgot to push the DMA change, so it didn't end up in
v5.4-surface. Thanks for noticing! Should be fixed now.Also thanks for checking!
drm/kfd: fix a system crash issue during GPU recovery
drm/kfd: fix a system crash issue during GPU recovery
net: ftgmac100: Fix crash when removing driver
net: ftgmac100: Fix crash when removing driver
net: ftgmac100: Fix crash when removing driver
net: ftgmac100: Fix crash when removing driver
selftests/bpf: Fix BPF_CORE_READ_BITFIELD() macro
selftests/bpf: Fix BPF_CORE_READ_BITFIELD() macro
selftests/bpf: Fix BPF_CORE_READ_BITFIELD() macro
um: Mark all kernel symbols as local
um: Mark all kernel symbols as local
um: Mark all kernel symbols as local
powerpc/stacktrace: Fix spurious "stale" traces in raise_backtrace_ipi()
powerpc/stacktrace: Fix spurious "stale" traces in raise_backtrace_ipi()
powerpc/stacktrace: Fix spurious "stale" traces in raise_backtrace_ipi()
kernel/locking: Add context to ww_mutex_trylock()
ipmi: Fix UAF when uninstall ipmi_si and ipmi_msghandler module
ipmi: Fix UAF when uninstall ipmi_si and ipmi_msghandler module
drm/amdgpu: Fix a null pointer access when the smc_rreg pointer is NULL
drm/amdgpu: Fix a null pointer access when the smc_rreg pointer is NULL