Issue with JFFS2 and a_ops->dirty_folio
Jean-Michel Hautbois
jeanmichel.hautbois at yoseli.org
Thu Jun 13 05:57:50 PDT 2024
Hello Christoph,
On 13/06/2024 13:20, Christoph Hellwig wrote:
> On Thu, Jun 13, 2024 at 09:05:17AM +0200, Jean-Michel Hautbois wrote:
>> Hi everyone !
>>
>> I am currently working on a Coldfire (MPC54418) and quite everything goes
>> well, except that I can only execute one command from user space before
>> getting a segmentation fault on the do_exit() syscall.
>
> Looks like jffs2 is simply missing a dirty_folio implementation. The
> simple filemap_dirty_folio should do the job, please try the patch
> below:
>
>
> diff --git a/fs/jffs2/file.c b/fs/jffs2/file.c
> index 62ea76da7fdf23..7124cbad6c35ae 100644
> --- a/fs/jffs2/file.c
> +++ b/fs/jffs2/file.c
> @@ -19,6 +19,7 @@
> #include <linux/highmem.h>
> #include <linux/crc32.h>
> #include <linux/jffs2.h>
> +#include <linux/writeback.h>
> #include "nodelist.h"
>
> static int jffs2_write_end(struct file *filp, struct address_space *mapping,
> @@ -75,6 +76,7 @@ const struct address_space_operations jffs2_file_address_operations =
> .read_folio = jffs2_read_folio,
> .write_begin = jffs2_write_begin,
> .write_end = jffs2_write_end,
> + .dirty_folio = filemap_dirty_folio,
> };
>
> static int jffs2_do_readpage_nolock (struct inode *inode, struct page *pg)
Thanks, I did implement this one, but now I have another weird issue, I
don't know if this can be related...
When the bash command is launched (my init command is init=/bin/bash) I
can launch a first command (say, ls for instance) and it works fine. But
a second call to this same command or any other one juste returns as if
nothing was done... And I can't even debug, strace fails too:
execve("/bin/ls", ["/bin/ls"], 0xbfb31ef0 /* 5 vars */) = 0
brk(NULL) = 0x2ab7c000
atomic_barrier() = 0
access("/etc/ld.so.preload", R_OK) = -1 ENOENT (No such file or
directory)
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
openat(AT_FDCWD, "/etc/ld.so.cache", O_RDONLY|O_LARGEFILE|O_CLOEXEC) =
-1 ENOENT (No such file or directory)
openat(AT_FDCWD, "/lib/libresolv.so.2", O_RDONLY|O_LARGEFILE|O_CLOEXEC) = 3
read(3,
"\177ELF\1\2\1\0\0\0\0\0\0\0\0\0\0\3\0\4\0\0\0\1\0\0\0\0\0\0\0004"...,
512) = 512
statx(3, "", AT_STATX_SYNC_AS_STAT|AT_NO_AUTOMOUNT|AT_EMPTY_PATH,
STATX_BASIC_STATS,
{stx_mask=STATX_TYPE|STATX_MODE|STATX_NLINK|STATX_UID|STATX_GID|STATX_MTIME|STATX_CTIME|STATX_INO|STATX_SIZE|STATX_BLOCKS|STATX_
MNT_ID, stx_attributes=0, stx_mode=S_IFREG|0755, stx_size=43120, ...}) = 0
mmap2(NULL, 59888, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_DENYWRITE, 3, 0)
=[ 15.830000] random: crng init done
0x60022000
mmap2(0x6002c000, 16384, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_FIXED|MAP_DENYWRITE, 3, 0x8000) = 0x6002c000
mmap2(0x60030000, 2544, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS, -1, 0) = 0x60030000
close(3) = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
openat(AT_FDCWD, "/lib/libc.so.6", O_RDONLY|O_LARGEFILE|O_CLOEXEC) = 3
read(3,
"\177ELF\1\2\1\0\0\0\0\0\0\0\0\0\0\3\0\4\0\0\0\1\0\2\324\n\0\0\0004"...,
512) = 512
statx(3, "", AT_STATX_SYNC_AS_STAT|AT_NO_AUTOMOUNT|AT_EMPTY_PATH,
STATX_BASIC_STATS,
{stx_mask=STATX_TYPE|STATX_MODE|STATX_NLINK|STATX_UID|STATX_GID|STATX_MTIME|STATX_CTIME|STATX_INO|STATX_SIZE|STATX_BLOCKS|STATX_
MNT_ID, stx_attributes=0, stx_mode=S_IFREG|0755, stx_size=1257660, ...}) = 0
mmap2(NULL, 1290920, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_DENYWRITE, 3,
0) = 0x60032000
mmap2(0x6015e000, 24576, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_FIXED|MAP_DENYWRITE, 3, 0x12c000) = 0x6015e000
mmap2(0x60164000, 37544, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS, -1, 0) = 0x60164000
close(3) = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
atomic_barrier() = 0
mmap2(NULL, 16384, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1,
0) = 0x6016e000
set_thread_area(0x601759c0) = 0
get_thread_area() = 0x601759c0
atomic_barrier() = 0
set_tid_address(0x6016e548) = 28
set_robust_list(0x6016e54c, 12) = 0
mprotect(0x6015e000, 8192, PROT_READ) = 0
mprotect(0x6002c000, 8192, PROT_READ) = 0
mprotect(0x2ab72000, 8192, PROT_READ) = 0
mprotect(0x6001e000, 8192, PROT_READ) = 0
--- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=0xc00815f4} ---
+++ killed by SIGSEGV +++
I suppose this can be related to the ELF_DT_DYN_BASE address, but I
can't see what is going on yet.
Thanks,
JM
More information about the linux-mtd
mailing list