Message ID | 1432caf4ecfb077bb54f0115cb5e12b6620e8bfa.1515439558.git.robin.murphy@arm.com (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
Hi Robin, On 2018-01-08 20:27, Robin Murphy wrote: > Removing the early device registration hook overlooked the fact that > it only ran conditionally on a compatible device being present in the > DT. With exynos_iommu_init() now running as an unconditional initcall, > problems arise on non-Exynos systems when other IOMMU drivers find > themselves unable to install their ops on the platform bus, or at worst > the Exynos ops get called with someone else's domain and all hell breaks > loose. > > Fix this by delaying the setting of bus ops until an Exynos IOMMU is > actually found, to replicate the previous order of events. > > Fixes: 928055a01b3f ("iommu/exynos: Remove custom platform device registration code") > Signed-off-by: Robin Murphy <robin.murphy@arm.com> Right, my fault. However I will prefer to resurrect code added initially by commit a7b67cd5d9af "iommu/exynos: Play nice in multi-platform builds". There is no need to do all the things done in the exynos_iommu_init on non-Exynos platforms. > --- > drivers/iommu/exynos-iommu.c | 16 +++++++--------- > 1 file changed, 7 insertions(+), 9 deletions(-) > > diff --git a/drivers/iommu/exynos-iommu.c b/drivers/iommu/exynos-iommu.c > index 6a96a4c42153..e9e756156429 100644 > --- a/drivers/iommu/exynos-iommu.c > +++ b/drivers/iommu/exynos-iommu.c > @@ -574,6 +574,12 @@ static int __init exynos_sysmmu_probe(struct platform_device *pdev) > struct sysmmu_drvdata *data; > struct resource *res; > > + if (platform_bus_type->iommu_ops != &exynos_iommu_ops) { > + ret = bus_set_iommu(&platform_bus_type, &exynos_iommu_ops); > + if (ret) > + return ret; > + } > + > data = devm_kzalloc(dev, sizeof(*data), GFP_KERNEL); > if (!data) > return -ENOMEM; > @@ -1367,16 +1373,8 @@ static int __init exynos_iommu_init(void) > goto err_zero_lv2; > } > > - ret = bus_set_iommu(&platform_bus_type, &exynos_iommu_ops); > - if (ret) { > - pr_err("%s: Failed to register exynos-iommu driver.\n", > - __func__); > - goto err_set_iommu; > - } > - > return 0; > -err_set_iommu: > - kmem_cache_free(lv2table_kmem_cache, zero_lv2_table); > + > err_zero_lv2: > platform_driver_unregister(&exynos_sysmmu_driver); > err_reg_driver: Best regards
On 09/01/18 09:59, Marek Szyprowski wrote: > Hi Robin, > > On 2018-01-08 20:27, Robin Murphy wrote: >> Removing the early device registration hook overlooked the fact that >> it only ran conditionally on a compatible device being present in the >> DT. With exynos_iommu_init() now running as an unconditional initcall, >> problems arise on non-Exynos systems when other IOMMU drivers find >> themselves unable to install their ops on the platform bus, or at worst >> the Exynos ops get called with someone else's domain and all hell breaks >> loose. >> >> Fix this by delaying the setting of bus ops until an Exynos IOMMU is >> actually found, to replicate the previous order of events. >> >> Fixes: 928055a01b3f ("iommu/exynos: Remove custom platform device >> registration code") >> Signed-off-by: Robin Murphy <robin.murphy@arm.com> > > Right, my fault. However I will prefer to resurrect code added initially > by commit a7b67cd5d9af "iommu/exynos: Play nice in multi-platform builds". > There is no need to do all the things done in the exynos_iommu_init on > non-Exynos platforms. Yeah; I had a moment of doubt and left the rest as-is, but I guess all of that global setup could in fact be delayed until the first probe, as with dma_dev. Anyway, I'll respin just the minimal fix to un-break multiplatform, and leave any further refactoring up to you. Thanks, Robin. >> --- >> drivers/iommu/exynos-iommu.c | 16 +++++++--------- >> 1 file changed, 7 insertions(+), 9 deletions(-) >> >> diff --git a/drivers/iommu/exynos-iommu.c b/drivers/iommu/exynos-iommu.c >> index 6a96a4c42153..e9e756156429 100644 >> --- a/drivers/iommu/exynos-iommu.c >> +++ b/drivers/iommu/exynos-iommu.c >> @@ -574,6 +574,12 @@ static int __init exynos_sysmmu_probe(struct >> platform_device *pdev) >> struct sysmmu_drvdata *data; >> struct resource *res; >> + if (platform_bus_type->iommu_ops != &exynos_iommu_ops) { >> + ret = bus_set_iommu(&platform_bus_type, &exynos_iommu_ops); >> + if (ret) >> + return ret; >> + } >> + >> data = devm_kzalloc(dev, sizeof(*data), GFP_KERNEL); >> if (!data) >> return -ENOMEM; >> @@ -1367,16 +1373,8 @@ static int __init exynos_iommu_init(void) >> goto err_zero_lv2; >> } >> - ret = bus_set_iommu(&platform_bus_type, &exynos_iommu_ops); >> - if (ret) { >> - pr_err("%s: Failed to register exynos-iommu driver.\n", >> - __func__); >> - goto err_set_iommu; >> - } >> - >> return 0; >> -err_set_iommu: >> - kmem_cache_free(lv2table_kmem_cache, zero_lv2_table); >> + >> err_zero_lv2: >> platform_driver_unregister(&exynos_sysmmu_driver); >> err_reg_driver: > > Best regards
diff --git a/drivers/iommu/exynos-iommu.c b/drivers/iommu/exynos-iommu.c index 6a96a4c42153..e9e756156429 100644 --- a/drivers/iommu/exynos-iommu.c +++ b/drivers/iommu/exynos-iommu.c @@ -574,6 +574,12 @@ static int __init exynos_sysmmu_probe(struct platform_device *pdev) struct sysmmu_drvdata *data; struct resource *res; + if (platform_bus_type->iommu_ops != &exynos_iommu_ops) { + ret = bus_set_iommu(&platform_bus_type, &exynos_iommu_ops); + if (ret) + return ret; + } + data = devm_kzalloc(dev, sizeof(*data), GFP_KERNEL); if (!data) return -ENOMEM; @@ -1367,16 +1373,8 @@ static int __init exynos_iommu_init(void) goto err_zero_lv2; } - ret = bus_set_iommu(&platform_bus_type, &exynos_iommu_ops); - if (ret) { - pr_err("%s: Failed to register exynos-iommu driver.\n", - __func__); - goto err_set_iommu; - } - return 0; -err_set_iommu: - kmem_cache_free(lv2table_kmem_cache, zero_lv2_table); + err_zero_lv2: platform_driver_unregister(&exynos_sysmmu_driver); err_reg_driver:
Removing the early device registration hook overlooked the fact that it only ran conditionally on a compatible device being present in the DT. With exynos_iommu_init() now running as an unconditional initcall, problems arise on non-Exynos systems when other IOMMU drivers find themselves unable to install their ops on the platform bus, or at worst the Exynos ops get called with someone else's domain and all hell breaks loose. Fix this by delaying the setting of bus ops until an Exynos IOMMU is actually found, to replicate the previous order of events. Fixes: 928055a01b3f ("iommu/exynos: Remove custom platform device registration code") Signed-off-by: Robin Murphy <robin.murphy@arm.com> --- drivers/iommu/exynos-iommu.c | 16 +++++++--------- 1 file changed, 7 insertions(+), 9 deletions(-)