| Message ID | 20181113184910.26697-1-mhocko@kernel.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | l1tf: drop the swap storage limit restriction when l1tf=off | expand |
On 18-11-13 19:49:10, Michal Hocko wrote: > From: Michal Hocko <mhocko@suse.com> > > Swap storage is restricted to max_swapfile_size (~16TB on x86_64) > whenever the system is deemed affected by L1TF vulnerability. Even > though the limit is quite high for most deployments it seems to be > too restrictive for deployments which are willing to live with the > mitigation disabled. > > We have a customer to deploy 8x 6,4TB PCIe/NVMe SSD swap devices > which is clearly out of the limit. > > Drop the swap restriction when l1tf=off is specified. It also doesn't > make much sense to warn about too much memory for the l1tf mitigation > when it is forcefully disabled by the administrator. > > Signed-off-by: Michal Hocko <mhocko@suse.com> Reviewed-by: Pavel Tatashin <pasha.tatashin@soleen.com>
On Tue, 13 Nov 2018, Michal Hocko wrote: > From: Michal Hocko <mhocko@suse.com> > > Swap storage is restricted to max_swapfile_size (~16TB on x86_64) > whenever the system is deemed affected by L1TF vulnerability. Even > though the limit is quite high for most deployments it seems to be > too restrictive for deployments which are willing to live with the > mitigation disabled. > > We have a customer to deploy 8x 6,4TB PCIe/NVMe SSD swap devices > which is clearly out of the limit. > > Drop the swap restriction when l1tf=off is specified. It also doesn't > make much sense to warn about too much memory for the l1tf mitigation > when it is forcefully disabled by the administrator. > > Signed-off-by: Michal Hocko <mhocko@suse.com> > --- > Documentation/admin-guide/kernel-parameters.txt | 2 ++ > Documentation/admin-guide/l1tf.rst | 5 ++++- > arch/x86/kernel/cpu/bugs.c | 3 ++- > arch/x86/mm/init.c | 2 +- > 4 files changed, 9 insertions(+), 3 deletions(-) > > diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt > index 81d1d5a74728..a54f2bd39e77 100644 > --- a/Documentation/admin-guide/kernel-parameters.txt > +++ b/Documentation/admin-guide/kernel-parameters.txt > @@ -2095,6 +2095,8 @@ > off > Disables hypervisor mitigations and doesn't > emit any warnings. > + It also drops the swap size and available > + RAM limit restriction. Minor nit: I think this should explicitly mention that those two things are related to bare metal mitigation, to avoid any confusion (as otherwise the l1tf cmdline parameter is purely about hypervisor mitigations). With that Acked-by: Jiri Kosina <jkosina@suse.cz> Thanks,
On Tue 13-11-18 20:56:54, Jiri Kosina wrote: > On Tue, 13 Nov 2018, Michal Hocko wrote: > > > From: Michal Hocko <mhocko@suse.com> > > > > Swap storage is restricted to max_swapfile_size (~16TB on x86_64) > > whenever the system is deemed affected by L1TF vulnerability. Even > > though the limit is quite high for most deployments it seems to be > > too restrictive for deployments which are willing to live with the > > mitigation disabled. > > > > We have a customer to deploy 8x 6,4TB PCIe/NVMe SSD swap devices > > which is clearly out of the limit. > > > > Drop the swap restriction when l1tf=off is specified. It also doesn't > > make much sense to warn about too much memory for the l1tf mitigation > > when it is forcefully disabled by the administrator. > > > > Signed-off-by: Michal Hocko <mhocko@suse.com> > > --- > > Documentation/admin-guide/kernel-parameters.txt | 2 ++ > > Documentation/admin-guide/l1tf.rst | 5 ++++- > > arch/x86/kernel/cpu/bugs.c | 3 ++- > > arch/x86/mm/init.c | 2 +- > > 4 files changed, 9 insertions(+), 3 deletions(-) > > > > diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt > > index 81d1d5a74728..a54f2bd39e77 100644 > > --- a/Documentation/admin-guide/kernel-parameters.txt > > +++ b/Documentation/admin-guide/kernel-parameters.txt > > @@ -2095,6 +2095,8 @@ > > off > > Disables hypervisor mitigations and doesn't > > emit any warnings. > > + It also drops the swap size and available > > + RAM limit restriction. > > Minor nit: I think this should explicitly mention that those two things > are related to bare metal mitigation, to avoid any confusion (as otherwise > the l1tf cmdline parameter is purely about hypervisor mitigations). Do you have any specific wording in mind? It also drops the swap size and available RAM limit restrictions on both hypervisor and bare metal. Sounds better? > With that > > Acked-by: Jiri Kosina <jkosina@suse.cz> Thanks!
On Tue, Nov 13, 2018 at 07:49:10PM +0100, Michal Hocko wrote: > From: Michal Hocko <mhocko@suse.com> > > Swap storage is restricted to max_swapfile_size (~16TB on x86_64) > whenever the system is deemed affected by L1TF vulnerability. Even > though the limit is quite high for most deployments it seems to be > too restrictive for deployments which are willing to live with the > mitigation disabled. > > We have a customer to deploy 8x 6,4TB PCIe/NVMe SSD swap devices > which is clearly out of the limit. > > Drop the swap restriction when l1tf=off is specified. It also doesn't > make much sense to warn about too much memory for the l1tf mitigation > when it is forcefully disabled by the administrator. Reviewed-by: Andi Kleen <ak@linux.intel.com> -Andi
On Wed, 14 Nov 2018, Michal Hocko wrote: > > > + It also drops the swap size and available > > > + RAM limit restriction. > > > > Minor nit: I think this should explicitly mention that those two things > > are related to bare metal mitigation, to avoid any confusion (as otherwise > > the l1tf cmdline parameter is purely about hypervisor mitigations). > > Do you have any specific wording in mind? > > It also drops the swap size and available RAM limit restrictions on both > hypervisor and bare metal. > > Sounds better? > > > With that > > > > Acked-by: Jiri Kosina <jkosina@suse.cz> > > Thanks! Yes, I think that makes it absolutely clear. Thanks,
On Mon 19-11-18 14:36:32, Jiri Kosina wrote: > On Wed, 14 Nov 2018, Michal Hocko wrote: > > > > > + It also drops the swap size and available > > > > + RAM limit restriction. > > > > > > Minor nit: I think this should explicitly mention that those two things > > > are related to bare metal mitigation, to avoid any confusion (as otherwise > > > the l1tf cmdline parameter is purely about hypervisor mitigations). > > > > Do you have any specific wording in mind? > > > > It also drops the swap size and available RAM limit restrictions on both > > hypervisor and bare metal. > > > > Sounds better? > > > > > With that > > > > > > Acked-by: Jiri Kosina <jkosina@suse.cz> > > > > Thanks! > > Yes, I think that makes it absolutely clear. Thanks, OK. Here is the incremental diff on top of the patch. I will fold and repost later this week. I assume people are still catching up after LPC and I do not want to spam them even more. diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt index a54f2bd39e77..c5aa4b4a797d 100644 --- a/Documentation/admin-guide/kernel-parameters.txt +++ b/Documentation/admin-guide/kernel-parameters.txt @@ -2096,7 +2096,8 @@ Disables hypervisor mitigations and doesn't emit any warnings. It also drops the swap size and available - RAM limit restriction. + RAM limit restriction on both hypervisor and + bare metal. Default is 'flush'. diff --git a/Documentation/admin-guide/l1tf.rst b/Documentation/admin-guide/l1tf.rst index b00464a9c09c..2e65e6cb033e 100644 --- a/Documentation/admin-guide/l1tf.rst +++ b/Documentation/admin-guide/l1tf.rst @@ -405,7 +405,8 @@ The kernel command line allows to control the L1TF mitigations at boot off Disables hypervisor mitigations and doesn't emit any warnings. - It also drops the swap size and available RAM limit restrictions. + It also drops the swap size and available RAM limit restrictions + on both hypervisor and bare metal. ============ =============================================================
On Mon, 19 Nov 2018, Michal Hocko wrote: > > > > > + It also drops the swap size and available > > > > > + RAM limit restriction. > > > > > > > > Minor nit: I think this should explicitly mention that those two things > > > > are related to bare metal mitigation, to avoid any confusion (as otherwise > > > > the l1tf cmdline parameter is purely about hypervisor mitigations). > > > > > > Do you have any specific wording in mind? > > > > > > It also drops the swap size and available RAM limit restrictions on both > > > hypervisor and bare metal. > > > > > > Sounds better? > > > > > > > With that > > > > > > > > Acked-by: Jiri Kosina <jkosina@suse.cz> > > > > > > Thanks! > > > > Yes, I think that makes it absolutely clear. Thanks, > > OK. Here is the incremental diff on top of the patch. I will fold and > repost later this week. I assume people are still catching up after LPC > and I do not want to spam them even more. Is this queued anywhere in the meantime please? Thanks,
On Sun 09-12-18 21:46:57, Jiri Kosina wrote: > On Mon, 19 Nov 2018, Michal Hocko wrote: > > > > > > > + It also drops the swap size and available > > > > > > + RAM limit restriction. > > > > > > > > > > Minor nit: I think this should explicitly mention that those two things > > > > > are related to bare metal mitigation, to avoid any confusion (as otherwise > > > > > the l1tf cmdline parameter is purely about hypervisor mitigations). > > > > > > > > Do you have any specific wording in mind? > > > > > > > > It also drops the swap size and available RAM limit restrictions on both > > > > hypervisor and bare metal. > > > > > > > > Sounds better? > > > > > > > > > With that > > > > > > > > > > Acked-by: Jiri Kosina <jkosina@suse.cz> > > > > > > > > Thanks! > > > > > > Yes, I think that makes it absolutely clear. Thanks, > > > > OK. Here is the incremental diff on top of the patch. I will fold and > > repost later this week. I assume people are still catching up after LPC > > and I do not want to spam them even more. > > Is this queued anywhere in the meantime please? Not yet. Thanks for the reminder. It completely fall of my radar. Thomas, do you want me to resubmit or there are some other changes you would like to see?
diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt index 81d1d5a74728..a54f2bd39e77 100644 --- a/Documentation/admin-guide/kernel-parameters.txt +++ b/Documentation/admin-guide/kernel-parameters.txt @@ -2095,6 +2095,8 @@ off Disables hypervisor mitigations and doesn't emit any warnings. + It also drops the swap size and available + RAM limit restriction. Default is 'flush'. diff --git a/Documentation/admin-guide/l1tf.rst b/Documentation/admin-guide/l1tf.rst index b85dd80510b0..b00464a9c09c 100644 --- a/Documentation/admin-guide/l1tf.rst +++ b/Documentation/admin-guide/l1tf.rst @@ -405,6 +405,8 @@ The kernel command line allows to control the L1TF mitigations at boot off Disables hypervisor mitigations and doesn't emit any warnings. + It also drops the swap size and available RAM limit restrictions. + ============ ============================================================= The default is 'flush'. For details about L1D flushing see :ref:`l1d_flush`. @@ -576,7 +578,8 @@ Default mitigations The kernel default mitigations for vulnerable processors are: - PTE inversion to protect against malicious user space. This is done - unconditionally and cannot be controlled. + unconditionally and cannot be controlled. The swap storage is limited + to ~16TB. - L1D conditional flushing on VMENTER when EPT is enabled for a guest. diff --git a/arch/x86/kernel/cpu/bugs.c b/arch/x86/kernel/cpu/bugs.c index c37e66e493bf..761100cd3eab 100644 --- a/arch/x86/kernel/cpu/bugs.c +++ b/arch/x86/kernel/cpu/bugs.c @@ -779,7 +779,8 @@ static void __init l1tf_select_mitigation(void) #endif half_pa = (u64)l1tf_pfn_limit() << PAGE_SHIFT; - if (e820__mapped_any(half_pa, ULLONG_MAX - half_pa, E820_TYPE_RAM)) { + if (l1tf_mitigation != L1TF_MITIGATION_OFF && + e820__mapped_any(half_pa, ULLONG_MAX - half_pa, E820_TYPE_RAM)) { pr_warn("System has more than MAX_PA/2 memory. L1TF mitigation not effective.\n"); pr_info("You may make it effective by booting the kernel with mem=%llu parameter.\n", half_pa); diff --git a/arch/x86/mm/init.c b/arch/x86/mm/init.c index ef99f3892e1f..427a955a2cf2 100644 --- a/arch/x86/mm/init.c +++ b/arch/x86/mm/init.c @@ -931,7 +931,7 @@ unsigned long max_swapfile_size(void) pages = generic_max_swapfile_size(); - if (boot_cpu_has_bug(X86_BUG_L1TF)) { + if (boot_cpu_has_bug(X86_BUG_L1TF) && l1tf_mitigation != L1TF_MITIGATION_OFF) { /* Limit the swap file size to MAX_PA/2 for L1TF workaround */ unsigned long long l1tf_limit = l1tf_pfn_limit(); /*