Diffusion FreeBSD src repository 70f51f0e474f

Revert "Handle partial reads in zfs_read"
70f51f0e474f
Actions

Description

Revert "Handle partial reads in zfs_read"

This reverts commit 59eab1093a361ca76849ca76b2ec079c8736e8e3.

The change suppressed EFAULT originating from uiomove(). The deadlock
avoidance mechanism implemented by vn_io_fault1() in the VFS handles
such errors by wiring the user pages and retrying, but this change
caused read() to return early instead. This can result in short I/O,
causing misbehaviour in some applications, and possibly other
consequences.

Until this is resolved somehow, revert the commit.

Approved by: mm

Details

Provenance

markj

Authored on Oct 22 2021, 6:55 PM

Parents

rG6aae3517ed25: Retire synchronous PPP kernel driver sppp(4).

Branches

Unknown

Tags

Unknown

Reverts

rG59eab1093a36: Handle partial reads in zfs_read

Event Timeline

markj committed rG70f51f0e474f: Revert "Handle partial reads in zfs_read" (authored by markj).Oct 22 2021, 7:16 PM

markj added a reverted change: rG59eab1093a36: Handle partial reads in zfs_read.Oct 22 2021, 7:20 PM

Forgive me if there's an obvious link on Phabricator I'm overlooking, but is there a testcase or bug report that demonstrates how this is broken? I'd like to figure out where this behavior is and isn't wrong, and hopefully revise the patch...

In R10:70f51f0e474ffe1fb74cb427423a2fba3637544d#412065, @rincebrain_gmail.com wrote:

Forgive me if there's an obvious link on Phabricator I'm overlooking, but is there a testcase or bug report that demonstrates how this is broken? I'd like to figure out where this behavior is and isn't wrong, and hopefully revise the patch...

You're not overlooking anything. I don't have a reduced test case yet. I just saw some toolchain program crashes resulting from mishandling of short reads, and a couple of reports of similar problems on the mailing lists. I'll try and narrow it down this weekend so I can submit a proper bug report.

In R10:70f51f0e474ffe1fb74cb427423a2fba3637544d#412068, @markj wrote:

In R10:70f51f0e474ffe1fb74cb427423a2fba3637544d#412065, @rincebrain_gmail.com wrote:

Forgive me if there's an obvious link on Phabricator I'm overlooking, but is there a testcase or bug report that demonstrates how this is broken? I'd like to figure out where this behavior is and isn't wrong, and hopefully revise the patch...

You're not overlooking anything. I don't have a reduced test case yet. I just saw some toolchain program crashes resulting from mishandling of short reads, and a couple of reports of similar problems on the mailing lists. I'll try and narrow it down this weekend so I can submit a proper bug report.