Re: Filesystem errors on armv7 -current

From: Mark Millard <marklmi_at_yahoo.com>
Date: Wed, 21 Oct 2026 19:24:47 UTC
On 10/21/26 11:33, bob prohaska wrote:
> On Mon, Oct 19, 2026 at 06:43:59PM -0700, bob prohaska wrote:
>> On Mon, Oct 19, 2026 at 11:47:56AM -0700, Mark Millard wrote:
>>>
>>> What about any newer security Email reports? Are they still showing find
>>> reporting the various "Input/output error" messages? In the published
>>> file they repeated for 2 distinct find commands.
> 
> The security email this morning (10/21/26) contained in part:
> 
> 
> Checking setuid files and devices:
> find: /usr/ports/accessibility/py-atspi: Input/output error
> find: /usr/ports/biology/smithwaterman: Input/output error
> find: /usr/ports/devel/dwz/files: Input/output error
> find: /usr/ports/devel/p5-Config-Parser: Input/output error
> find: /usr/src/lib/libc/resolv: Input/output error
> find: /usr/src/lib/libpmc/pmu-events/arch/arm64/ampere: Input/output error
> find: /usr/src/sys/contrib/device-tree/Bindings/powerpc/nintendo: Input/output
> +error
> find: /usr/src/sys/contrib/device-tree/Bindings/reset: Input/output error
> find:
> +/usr/src/sys/contrib/libsodium/src/libsodium/crypto_stream/salsa20/xmm6int:
> +Input/output error
> 
> 
> Checking negative group permissions:
> find: /usr/ports/accessibility/py-atspi: Input/output error
> find: /usr/ports/biology/smithwaterman: Input/output error
> find: /usr/ports/devel/dwz/files: Input/output error
> find: /usr/ports/devel/p5-Config-Parser: Input/output error
> find: /usr/src/lib/libc/resolv: Input/output error
> find: /usr/src/lib/libpmc/pmu-events/arch/arm64/ampere: Input/output error
> find: /usr/src/sys/contrib/device-tree/Bindings/powerpc/nintendo: Input/output
> +error
> find: /usr/src/sys/contrib/device-tree/Bindings/reset: Input/output error
> find:
> +/usr/src/sys/contrib/libsodium/src/libsodium/crypto_stream/salsa20/xmm6int:
> +Input/output error
> 
> However, buildworld/buildkernel and installation of both succeeded last
> night, followed by a successful hands-off reboot. A re-run of buildworld
> on the previously-built sources is underway now to get /usr/obj up to date.

Just because your media likely has failed, does not mean that it has
failed everywhere all at once.

I have no clue how many of your lost+found items were formerly
directories or files that were getting EIO errors.

> 
> Running
> # ls -l /usr/ports/accessibility/py-atspi
> ls: /usr/ports/accessibility/py-atspi: Input/output error

The above likely indicates that some of the directory representation is
on part(s) of the media that are getting the errors. It need not involve
the file content.

> # 
> but
> # ls -ld /usr/ports/accessibility/py*
> drwxr-xr-x  2 root wheel 512 Dec  9  2025 /usr/ports/accessibility/py-atspi
> drwxr-xr-x  2 root wheel 512 Dec  9  2025 /usr/ports/accessibility/py-speech-dispatcher
> # 
> does not
> 
> Prior to the last OS build/install cycle /usr/src was renamed /usr/src.old
> and replaced by a fresh clone.
> 
> /usr/ports has been kept in its pre-existing form apart from git pulls
> 
> Does this maybe suggest a problem with find?

As far as I can tell, it suggests reliability problems with your media
in very specific places that repeatedly get the problem.

/usr/src/lib/libc/gen/errlst.c has:

const char *const sys_errlist[] = {
. . .
        "Input/output error",                   /*  5 - EIO */
. . .

/usr/src/lib/libc/include/errlst.h has:

#ifdef PIC
/* If the main executable imports these, do not use its copy from
libc.so. */
extern const char *const __hidden_sys_errlist[] __hidden;
extern const int __hidden_sys_nerr __hidden;
#else
#define __hidden_sys_errlist sys_errlist
#define __hidden_sys_nerr sys_nerr
#endif

And there are:

/usr/src/lib/libc/string/strerror.c:#include "errlst.h"
/usr/src/lib/libc/stdio/xprintf_errno.c:#include "errlst.h"
/usr/src/lib/libc/gen/errlst.c:#include "errlst.h"

find has:

# grep -r -e strerror -e xprintf_errno /usr/src/usr.bin/find/
/usr/src/usr.bin/find/find.c:			    entry->fts_path,
strerror(entry->fts_errno));

So find is reporting the EIO returned by the system.

> 
> Thanks for reading!
> 
> bob prohaska
> 
> 
> 
> 


-- 
===
Mark Millard
marklmi at yahoo.com