From: Benjamin Berg
In time-travel mode userspace can do a lot of work without any time
passing. Unfortunately, this can result in OOM situations as the RCU
core code will never be run.
Work around this by keeping track of userspace processes that do not
yield for a lot of operations. When this
Hi Mike,
On Mon, 9 Sept 2024 at 08:51, Mike Rapoport wrote:
>
> From: "Mike Rapoport (Microsoft)"
>
> Using large pages to map text areas reduces iTLB pressure and improves
> performance.
>
> Extend execmem_alloc() with an ability to use huge pages with ROX
> permissions as a cache for smaller a
Highmem was only supported on UML/i386. And the support has been
removed by commit a98a6d864d3b ("um: Remove broken highmem support").
Remove the leftovers and stop UML from trying to setup highmem when
the sum of physmem_size and iomem_size exceeds max_physmem.
Signed-off-by: Tiwei Bie
---
arch
Tiwei Bie (3):
um: Remove the redundant declaration of high_physmem
um: Remove highmem leftovers
um: Fix the definition for physmem_size
arch/um/drivers/virtio_uml.c | 9 -
arch/um/include/shared/as-layout.h | 3 ++-
arch/um/include/shared/mem_user.h | 5 ++---
arch/um/ker
Currently physmem_size is defined as long long but declared locally
as unsigned long long before using it in separate .c files. Make them
match by defining physmem_size as unsigned long long and also move
the declaration to a common header to allow the compiler to check it.
Signed-off-by: Tiwei Bi
high_physmem has already been declared in as-layout.h, so there is
no need to declare it explicitly in the .c file again.
While at it, group the declarations of __real_malloc and __real_free
together to make the code slightly more readable.
Signed-off-by: Tiwei Bie
---
arch/um/os-Linux/main.c |
From: Benjamin Berg
The function is not used anywhere.
Signed-off-by: Benjamin Berg
---
arch/um/include/shared/os.h | 1 -
arch/um/os-Linux/process.c | 5 -
2 files changed, 6 deletions(-)
diff --git a/arch/um/include/shared/os.h b/arch/um/include/shared/os.h
index a94093bfa5e4..e54f64f5
From: Benjamin Berg
The x86 port of UM has efficient unaligned access. Set the option as it
is appropriate and will e.g. cause UBSAN to not enable unaligned memory
access checking by default.
Signed-off-by: Benjamin Berg
---
arch/x86/um/Kconfig | 1 +
1 file changed, 1 insertion(+)
diff --git
From: Benjamin Berg
The function is not used anywhere.
Signed-off-by: Benjamin Berg
---
arch/um/include/shared/os.h | 1 -
arch/um/os-Linux/process.c | 5 -
2 files changed, 6 deletions(-)
diff --git a/arch/um/include/shared/os.h b/arch/um/include/shared/os.h
index 4bdd4fb5dd80..a94093bf
From: Benjamin Berg
The function is not used anywhere.
Signed-off-by: Benjamin Berg
---
arch/um/include/shared/os.h | 1 -
arch/um/os-Linux/process.c | 39 -
2 files changed, 40 deletions(-)
diff --git a/arch/um/include/shared/os.h b/arch/um/include/share
From: Benjamin Berg
This removes a set of unused functions and also selects
HAVE_EFFICIENT_UNALIGNED_ACCESS on x86 to avoid UBSAN spewing warnings
by default.
v2: Add Signed-off-by to all patches
Benjamin Berg (5):
um: Remove unused os_process_pc
um: Remove unused os_process_parent
um: Re
From: Benjamin Berg
The function is not used anywhere in the codebase.
Signed-off-by: Benjamin Berg
---
arch/um/include/shared/os.h | 1 -
arch/um/os-Linux/process.c | 33 -
2 files changed, 34 deletions(-)
diff --git a/arch/um/include/shared/os.h b/arch/um/i
From: Benjamin Berg
When switching from userspace to the kernel, all registers including the
FP registers are copied into the kernel and restored later on. As such,
the true source for the FP register state is actually already in the
kernel and they should never be grabbed from the userspace proc
From: Benjamin Berg
The x86 port of UM has efficient unaligned access. Set the option as it
is appropriate and will e.g. cause UBSAN to not enable unaligned memory
access checking by default.
---
arch/x86/um/Kconfig | 1 +
1 file changed, 1 insertion(+)
diff --git a/arch/x86/um/Kconfig b/arch/x
From: Benjamin Berg
The function is not used anywhere.
---
arch/um/include/shared/os.h | 1 -
arch/um/os-Linux/process.c | 5 -
2 files changed, 6 deletions(-)
diff --git a/arch/um/include/shared/os.h b/arch/um/include/shared/os.h
index a94093bfa5e4..e54f64f55bb7 100644
--- a/arch/um/inclu
From: Benjamin Berg
The function is not used anywhere.
---
arch/um/include/shared/os.h | 1 -
arch/um/os-Linux/process.c | 5 -
2 files changed, 6 deletions(-)
diff --git a/arch/um/include/shared/os.h b/arch/um/include/shared/os.h
index 4bdd4fb5dd80..a94093bfa5e4 100644
--- a/arch/um/inclu
From: Benjamin Berg
This removes a set of unused functions and also select
HAVE_EFFICIENT_UNALIGNED_ACCESS on x86 to avoid UBSAN spewing warnings
by default.
Benjamin Berg (5):
um: Remove unused os_process_pc
um: Remove unused os_process_parent
um: Remove unused os_stop_process
um: Remov
From: Benjamin Berg
The function is not used anywhere.
---
arch/um/include/shared/os.h | 1 -
arch/um/os-Linux/process.c | 39 -
2 files changed, 40 deletions(-)
diff --git a/arch/um/include/shared/os.h b/arch/um/include/shared/os.h
index c8c1a93c8d2c..4bdd
From: Benjamin Berg
The function is not used anywhere in the codebase.
---
arch/um/include/shared/os.h | 1 -
arch/um/os-Linux/process.c | 33 -
2 files changed, 34 deletions(-)
diff --git a/arch/um/include/shared/os.h b/arch/um/include/shared/os.h
index 9a039d
Hi,
On Fri, 2024-09-13 at 21:09 +0800, Tiwei Bie wrote:
> On 2024/9/13 16:22, Benjamin Berg wrote:
> > From: Benjamin Berg
> > [SNIP]
> > diff --git a/arch/um/kernel/process.c b/arch/um/kernel/process.c
> > index be2856af6d4c..ad798d40f8a4 100644
> > --- a/arch/um/kernel/process.c
> > +++ b/arch/
On 2024/9/13 16:22, Benjamin Berg wrote:
> From: Benjamin Berg
>
> When switching from userspace to the kernel, all registers including the
> FP registers are copied into the kernel and restored later on. As such,
> the true source for the FP register state is actually already in the
> kernel and
Hi!
- Ursprüngliche Mail -
> Von: "Benjamin Berg"
> First, it doesn't seem like my patch actually works, so please do not
> merge it. It actually appears that tree RCU and tiny RCU (which are
> selected depending on the preemption setting) are behaving differently.
>
> So now I am wonder
Hi
First, it doesn't seem like my patch actually works, so please do not
merge it. It actually appears that tree RCU and tiny RCU (which are
selected depending on the preemption setting) are behaving differently.
So now I am wondering if I can come up with a hack that works for both.
On Fri, 202
- Ursprüngliche Mail -
> Von: "Benjamin Berg"
>> While I acknowledge that time-travel itself is a beautiful hack, I'd
>> like to keep the hacks
>> to keep it working minimal.
>> So, the problem here is that RCU callbacks never run and just pile up?
>
> Yes. A simple example of this is doi
Hi,
On Thu, 2024-09-12 at 21:02 +0200, Richard Weinberger wrote:
> On Fri, Aug 30, 2024 at 5:38 PM Benjamin Berg
> wrote:
> >
> > From: Benjamin Berg
> >
> > In time-travel mode userspace can do a lot of work without any time
> > passing. Unfortunately, this can result in OOM situations as the
On Fri, 2024-09-13 at 10:22 +0200, Benjamin Berg wrote:
> From: Benjamin Berg
>
> When switching from userspace to the kernel, all registers including the
> FP registers are copied into the kernel and restored later on. As such,
> the true source for the FP register state is actually already in t
From: Benjamin Berg
When switching from userspace to the kernel, all registers including the
FP registers are copied into the kernel and restored later on. As such,
the true source for the FP register state is actually already in the
kernel and they should never be grabbed from the userspace proc
27 matches
Mail list logo
