Date: Tue, 26 Nov 2024 15:38:04 +0200 From: Konstantin Belousov <kib@freebsd.org> To: Dimitry Andric <dim@freebsd.org> Cc: Dag-Erling =?utf-8?B?U23DuHJncmF2?= <des@freebsd.org>, Mark Millard <marklmi@yahoo.com>, "jah@freebsd.org" <jah@freebsd.org>, dougm@freebsd.org, Alan Somers <asomers@freebsd.org>, Mark Johnston <markj@freebsd.org>, FreeBSD Current <freebsd-current@freebsd.org>, Guido Falsi <mad@madpilot.net>, Yasuhiro Kimura <yasu@freebsd.org>, ports@freebsd.org Subject: Re: port binary dumping core on recent head in poudriere [tmpfs corruptions involving blocks of zeros that should not be all zeros] Message-ID: <Z0XPPKtlLTMYeJS-@kib.kiev.ua> In-Reply-To: <4AE5B316-D7EB-4290-8D52-7FBF244EA7A4@FreeBSD.org> References: <38658C0D-CA33-4010-BBE1-E68D253A3DF7@FreeBSD.org> <1004a753-9a3c-4aa2-bfa8-4a0c471fe3ea@madpilot.net> <D14FF56C-506F-4168-91BC-1F10937B943F@yahoo.com> <E77AF0C3-5210-41C7-B8B8-02A8E22DB23D@yahoo.com> <A2820AEA-AB92-425F-AE91-2AF9629B3020@yahoo.com> <0690CFB1-6A6D-4B63-916C-BAB7F6256000@yahoo.com> <3660625A-0EE8-40DA-A248-EC18C734718C@yahoo.com> <865xoa2t6f.fsf@ltc.des.dev> <69A2E921-F5E3-40D2-977D-0964EE27349A@FreeBSD.org> <4AE5B316-D7EB-4290-8D52-7FBF244EA7A4@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, Nov 26, 2024 at 01:58:19PM +0100, Dimitry Andric wrote:
> On 26 Nov 2024, at 13:32, Dimitry Andric <dim@FreeBSD.org> wrote:
> >
> > On 26 Nov 2024, at 11:19, Dag-Erling Smørgrav <des@FreeBSD.org> wrote:
> >>
> >> Mark Millard <marklmi@yahoo.com> writes:
> >>> From inside a bulk -i where I did a manual make command
> >>> after it built and installed libsass.so.1.0.0 . The
> >>> manual make produced a /wrkdirs/ :
> >>> [...]
> >>> So the original creation looks okay. But . . .
> >>> [...]
> >>> So: The later, staged copy is a bad copy. Both are in the
> >>> tmpfs. So copying to the staging area makes a corrupted
> >>> copy inside the same tmpfs. After that, further copies of
> >>> staging's bad copy can be expected to be messed up.
> >>
> >> This and the fact that it happens on 14 and 15 but not on 13 strongly
> >> suggests an issue wth `copy_file_range(2)`, since `install(1)` in 14 and
> >> 15 (but not in 13) now uses `copy_file_range(2)` if at all possible.
> >>
> >> My educated guess is that hole detection doesn't work reliably for files
> >> that have had holes filled while memory-mapped, so `copy_file_range(2)`
> >> thinks there is a hole where there isn't one and skips some of the data
> >> when `install(1)` uses it to copy the library from `${WRKSRC}` to
> >> `${STAGEDIR}`. This may or may not be specific to tmpfs.
> >>
> >> You may want to try applying the attached patch to your FreeBSD 14 and
> >> 15 jails. It prevents `cp(1)` and `install(1)` from trying to use
> >> `copy_file_range(2)`.
> >
> > Yes, tmpfs is indeed the culprit (or at least involved). I have had USE_TMPFS=localbase in my poudriere.conf for a long time, since otherwise my build machine would run out of memory very quickly, so I didn't encounter any issues.
> >
> > Now I changed it to USE_TMPFS=yes, rebuilt only textproc/libsass and textproc/sassc, and then after reinstalling those packages:
> >
> > $ /usr/local/bin/sassc
> > Segmentation fault
>
> And after applying Dag-Erling's patch to disable copy_file_range for cp and install, it works correctly again.
So indeed there might be an issue in tmpfs seeking for data. Could you try
the following?
commit f4b848946a131dab260b44eab2cfabceb82bee0c
Author: Konstantin Belousov <kib@FreeBSD.org>
Date: Tue Nov 26 15:34:56 2024 +0200
tmpfs: do not skip pages searching for data
If the iterator finds invalid page at the requested pindex in
swap_pager_seek_data(), the current code only looks at the swap blocks
to search for data. This is not correct, valid pages may appear at the
higher indexes still.
diff --git a/sys/vm/swap_pager.c b/sys/vm/swap_pager.c
index db925f4ae7f6..390b2c10d680 100644
--- a/sys/vm/swap_pager.c
+++ b/sys/vm/swap_pager.c
@@ -2503,12 +2503,9 @@ swap_pager_seek_data(vm_object_t object, vm_pindex_t pindex)
VM_OBJECT_ASSERT_RLOCKED(object);
vm_page_iter_init(&pages, object);
m = vm_page_iter_lookup_ge(&pages, pindex);
- if (m != NULL) {
- if (!vm_page_any_valid(m))
- m = NULL;
- else if (pages.index == pindex)
- return (pages.index);
- }
+ if (m != NULL && pages.index == pindex)
+ return (pages.index);
+
swblk_iter_init_only(&blks, object);
swap_index = swap_pager_iter_find_least(&blks, pindex);
if (swap_index == pindex)
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Z0XPPKtlLTMYeJS->