[RISC-V] Disable EnableWriteXorExecute by default on riscv64 architectue #103408
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
On riscv64 architectue any dotnet process that loads corossgened assembly without DOTNET_EnableWriteXorExecute=0 environment variable set crashes with "Segmentation fault" error. For example System.Private.CoreLib.dll is compiled with crossgen2 by default during build phase so this crash applies to any processes that loads it. Setting EnableWriteXorExecute=0 disables this options on riscv64 architecure and prevents these crashes. Co-authored-by: Dong-Heon Jung <clamp03@gmail.com>
dotnet-policy-service
bot
added
the
community-contribution
rzsc
changed the title
Bajtazar
approved these changes
Jun 13, 2024
sirntar
approved these changes
Jun 13, 2024
yurai007
approved these changes
Jun 13, 2024
These crashes occur on both QEMU and VisionFive2 board. |
Is this crash a bug that should be fixed? |
Yes, it is a bug. On Now, before we start any dotnet process that loads a crossgened assembly we need to disable W^X by exporting What's more, there are some corefx tests that spawn child process with an empty environment ( That's why we need that W^X functionality to be disabled by default in runtime configuration for Of course when W^X gets correctly handled on |
|
@rzsc what is the memory page size on the devices where it crashes? And do you happen to know where the crash occurs? |
No, sorry. I cannot answer your questions because I'm not familiar with this crossgen2 vs W^X problem. What I'm trying to accomplish here is to move the mechanism of disabling W^X on Thank you. |
RISC-V test results for qemu-prio1-checked (
|
This was referenced Jun 13, 2024
RISC-V test results for starfive-prio1-checked (
|
This test was failing before PR changes. We are aware of it and working on it. |
Why are the corefx tests not passing with env variable? Is it the same problem #100505 by any chance? It would be better to fix that instead. If you would like to disable |
…rch. Co-authored-by: Jan Kotas <jkotas@microsoft.com>
Yes, this is the same problem as in #100505.
Yes, it sounds like a good idea. Thank you. |
RISC-V test results for qemu-prio1-checked (
|
RISC-V test results for starfive-prio1-checked (
|
jkotas
approved these changes
Jun 14, 2024
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
On riscv64 architectue any dotnet process that loads corossgened assembly without DOTNET_EnableWriteXorExecute=0 environment variable set crashes with "Segmentation fault" error.
For example System.Private.CoreLib.dll is compiled with crossgen2 by default during build phase so this crash applies to any processes that loads it.
Setting EnableWriteXorExecute=0 disables this options on riscv64 architecure and prevents these crashes.
Co-authored-by: Dong-Heon Jung clamp03@gmail.com
Part of #84834, cc @dotnet/samsung