Hi,
you have in Linux 5.10.13-rc1:
"x86/entry: Emit a symbol for register restoring thunk"
While that discussion Boris and Peter recommended to remove unused code via:
"x86/entry: Remove put_ret_addr_in_rdi THUNK macro argument" ( upstream commit 0bab9cb2d980d7c075cffb9216155f7835237f98 )
OK, this has no CC:stable but I have both as a series in my local Git and both were git-pulled from [1]. What do you think?
Regards, - Sedat -
[1] https://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git/log/?h=x86_entry... [2] https://git.kernel.org/linus/0bab9cb2d980d7c075cffb9216155f7835237f98
On Mon, Feb 01, 2021 at 08:50:52PM +0100, Sedat Dilek wrote:
Hi,
you have in Linux 5.10.13-rc1:
"x86/entry: Emit a symbol for register restoring thunk"
While that discussion Boris and Peter recommended to remove unused code via:
"x86/entry: Remove put_ret_addr_in_rdi THUNK macro argument" ( upstream commit 0bab9cb2d980d7c075cffb9216155f7835237f98 )
OK, this has no CC:stable but I have both as a series in my local Git and both were git-pulled from [1]. What do you think?
What bug is this fixing that requires this in 5.10?
thanks,
greg k-h
On Tue, Feb 2, 2021 at 2:06 PM Greg Kroah-Hartman gregkh@linuxfoundation.org wrote:
On Mon, Feb 01, 2021 at 08:50:52PM +0100, Sedat Dilek wrote:
Hi,
you have in Linux 5.10.13-rc1:
"x86/entry: Emit a symbol for register restoring thunk"
While that discussion Boris and Peter recommended to remove unused code via:
"x86/entry: Remove put_ret_addr_in_rdi THUNK macro argument" ( upstream commit 0bab9cb2d980d7c075cffb9216155f7835237f98 )
OK, this has no CC:stable but I have both as a series in my local Git and both were git-pulled from [1]. What do you think?
What bug is this fixing that requires this in 5.10?
Commit 0bab9cb2d980d7c075cffb9216155f7835237f98 removed unused logic.
So-to-say: Fixes: 320100a5ffe5 ("x86/entry: Remove the TRACE_IRQS cruft")
The commit was first introduced with Linux v5.8-rc1:
$ git describe --contains 320100a5ffe5 v5.8-rc1~21^2~28
As Linux v5.10.y is an LTS IMHO I hoped it is worth removing unused code. You better know the rules for stable-linux, so I leave it to you, Greg.
- Sedat -
On Wed, Feb 03, 2021 at 10:21:56AM +0100, Sedat Dilek wrote:
On Tue, Feb 2, 2021 at 2:06 PM Greg Kroah-Hartman gregkh@linuxfoundation.org wrote:
On Mon, Feb 01, 2021 at 08:50:52PM +0100, Sedat Dilek wrote:
Hi,
you have in Linux 5.10.13-rc1:
"x86/entry: Emit a symbol for register restoring thunk"
While that discussion Boris and Peter recommended to remove unused code via:
"x86/entry: Remove put_ret_addr_in_rdi THUNK macro argument" ( upstream commit 0bab9cb2d980d7c075cffb9216155f7835237f98 )
OK, this has no CC:stable but I have both as a series in my local Git and both were git-pulled from [1]. What do you think?
What bug is this fixing that requires this in 5.10?
Commit 0bab9cb2d980d7c075cffb9216155f7835237f98 removed unused logic.
So-to-say: Fixes: 320100a5ffe5 ("x86/entry: Remove the TRACE_IRQS cruft")
The commit was first introduced with Linux v5.8-rc1:
$ git describe --contains 320100a5ffe5 v5.8-rc1~21^2~28
As Linux v5.10.y is an LTS IMHO I hoped it is worth removing unused code. You better know the rules for stable-linux, so I leave it to you, Greg.
The rules are listed here: https://www.kernel.org/doc/html/latest/process/stable-kernel-rules.html
So please let me know what this patch fixes based on that and I will be glad to merge it.
thanks,
greg k-h
On Wed, Feb 3, 2021 at 11:04 AM Greg Kroah-Hartman gregkh@linuxfoundation.org wrote:
On Wed, Feb 03, 2021 at 10:21:56AM +0100, Sedat Dilek wrote:
On Tue, Feb 2, 2021 at 2:06 PM Greg Kroah-Hartman gregkh@linuxfoundation.org wrote:
On Mon, Feb 01, 2021 at 08:50:52PM +0100, Sedat Dilek wrote:
Hi,
you have in Linux 5.10.13-rc1:
"x86/entry: Emit a symbol for register restoring thunk"
While that discussion Boris and Peter recommended to remove unused code via:
"x86/entry: Remove put_ret_addr_in_rdi THUNK macro argument" ( upstream commit 0bab9cb2d980d7c075cffb9216155f7835237f98 )
OK, this has no CC:stable but I have both as a series in my local Git and both were git-pulled from [1]. What do you think?
What bug is this fixing that requires this in 5.10?
Commit 0bab9cb2d980d7c075cffb9216155f7835237f98 removed unused logic.
So-to-say: Fixes: 320100a5ffe5 ("x86/entry: Remove the TRACE_IRQS cruft")
The commit was first introduced with Linux v5.8-rc1:
$ git describe --contains 320100a5ffe5 v5.8-rc1~21^2~28
As Linux v5.10.y is an LTS IMHO I hoped it is worth removing unused code. You better know the rules for stable-linux, so I leave it to you, Greg.
The rules are listed here: https://www.kernel.org/doc/html/latest/process/stable-kernel-rules.html
So please let me know what this patch fixes based on that and I will be glad to merge it.
[ CC Boris Peter & Nick ]
I looked at the rules. The mentioned patch looks like a nice "cleanup" to me. If you are strict with the rules... Let's hear the CCed folks.
Thanks for your/the feedback (in advance).
- Sedat -
On Wed, Feb 03, 2021 at 01:30:50PM +0100, Sedat Dilek wrote:
Let's hear the CCed folks.
I have said this a couple of times already that the stable@ trees pick up too many patches already and the reasoning for some of them is debatable. But that is my opinion.
These patches are not even close to being stable material. Also my opinion.
HTH.
linux-stable-mirror@lists.linaro.org
-
Borislav Petkov -
Greg Kroah-Hartman -
Sedat Dilek