From: Andrej Picej <andrej.picej@norik.com>
To: barebox@lists.infradead.org
Subject: Re: [PATCH] scripts/common: fix write_file when opened with mmap
Date: Mon, 31 Jan 2022 11:23:39 +0100 [thread overview]
Message-ID: <813ef0f4-d6d4-1ca4-0729-09db490b811e@norik.com> (raw)
In-Reply-To: <20220131092358.GX23490@pengutronix.de>
Hi Sascha,
On 31. 01. 22 10:23, Sascha Hauer wrote:
> Hi Andrej,
>
> On Fri, Jan 28, 2022 at 09:00:33AM +0100, Andrej Picej wrote:
>> Usage of bareboximd with -c option (Create checksum for FILE and write
>> it to the crc32 tag) was broken. Possibly by applying 2154de1cf36c
>> (bareboximd: Use mmap when possibly). The script fails with:
>>
>> $ ./scripts/bareboximd -c images/<barebox-image.img>
>> Cannot write to images/<barebox-mage.img>: Bad address
>> CRC: write crc token to images/<barebox-image.img> failed: -14
>> Bad address
>>
>> This has to do with the usage of "mmap" and "open" with O_TRUNC flag
>> which truncates the file length to 0. Writing to files fails with:
>> EFAULT (14) buf is outside your accessible address space.
>> Remove the truncate flag and truncate manually after writing the data.
>>
>> This fixes the bareboximd script, which is now again usable with -c
>> option.
>
> That's a very unfortunate bug :(
>
>>
>> Signed-off-by: Anze Lesnik <anze.lesnik@norik.com>
>> Signed-off-by: Andrej Picej <andrej.picej@norik.com>
>> ---
>> scripts/common.c | 7 ++++++-
>> 1 file changed, 6 insertions(+), 1 deletion(-)
>>
>> diff --git a/scripts/common.c b/scripts/common.c
>> index 3f8dcd2d38..331988d593 100644
>> --- a/scripts/common.c
>> +++ b/scripts/common.c
>> @@ -107,7 +107,7 @@ int write_file(const char *filename, const void *buf, size_t size)
>> int fd, ret = 0;
>> int now;
>>
>> - fd = open(filename, O_WRONLY | O_TRUNC | O_CREAT,
>
> I added a comment above this while applying to prevent people from
> "optimizing" this patch away.
>
>> + fd = open(filename, O_WRONLY | O_CREAT,
>> S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH);
>> if (fd < 0) {
>> fprintf(stderr, "Cannot open %s: %s\n", filename, strerror(errno));
>> @@ -126,6 +126,11 @@ int write_file(const char *filename, const void *buf, size_t size)
>> buf += now;
>> }
>>
>> + if (ftruncate(fd, now) < 0) {
>> + fprintf(stderr, "Cannot truncate file: %s", strerror(errno));
>> + ret = -errno;
>> + }
>
> truncating to 'now' is only correct when we were able to write all bytes
> at once. Otherwise we must truncate to the input size. I fixed this up
> to the patch below.
That makes sense. Good thing you caught that.
Thanks.
>
> Sascha
>
> -----------------------------8<----------------------------
>
> From 738601e1258c55953284ee10801b26b9977918c2 Mon Sep 17 00:00:00 2001
> From: Andrej Picej <andrej.picej@norik.com>
> Date: Fri, 28 Jan 2022 09:00:33 +0100
> Subject: [PATCH] scripts/common: fix write_file when opened with mmap
>
> Usage of bareboximd with -c option (Create checksum for FILE and write
> it to the crc32 tag) was broken. Possibly by applying 2154de1cf36c
> (bareboximd: Use mmap when possibly). The script fails with:
>
> $ ./scripts/bareboximd -c images/<barebox-image.img>
> Cannot write to images/<barebox-mage.img>: Bad address
> CRC: write crc token to images/<barebox-image.img> failed: -14
> Bad address
>
> This has to do with the usage of "mmap" and "open" with O_TRUNC flag
> which truncates the file length to 0. Writing to files fails with:
> EFAULT (14) buf is outside your accessible address space.
> Remove the truncate flag and truncate manually after writing the data.
>
> This fixes the bareboximd script, which is now again usable with -c
> option.
>
> Signed-off-by: Anze Lesnik <anze.lesnik@norik.com>
> Signed-off-by: Andrej Picej <andrej.picej@norik.com>
> Link: https://lore.barebox.org/20220128080033.167251-1-andrej.picej@norik.com
> Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
> ---
> scripts/common.c | 15 +++++++++++----
> 1 file changed, 11 insertions(+), 4 deletions(-)
>
> diff --git a/scripts/common.c b/scripts/common.c
> index b780b09941..154d6dffcb 100644
> --- a/scripts/common.c
> +++ b/scripts/common.c
> @@ -105,26 +105,33 @@ int write_file(const char *filename, const void *buf, size_t size)
> {
> int fd, ret = 0;
> int now;
> + size_t left = size;
>
> - fd = open(filename, O_WRONLY | O_TRUNC | O_CREAT,
> + /* The same file may be mmapped currently, so can't use O_TRUNC here */
> + fd = open(filename, O_WRONLY | O_CREAT,
> S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH);
> if (fd < 0) {
> fprintf(stderr, "Cannot open %s: %s\n", filename, strerror(errno));
> return -errno;
> }
>
> - while (size) {
> - now = write(fd, buf, size);
> + while (left) {
> + now = write(fd, buf, left);
> if (now < 0) {
> fprintf(stderr, "Cannot write to %s: %s\n", filename,
> strerror(errno));
> ret = -errno;
> goto out;
> }
> - size -= now;
> + left -= now;
> buf += now;
> }
>
> + if (ftruncate(fd, size) < 0) {
> + fprintf(stderr, "Cannot truncate file: %s", strerror(errno));
> + ret = -errno;
> + }
> +
> out:
> close(fd);
>
>
_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox
prev parent reply other threads:[~2022-01-31 10:25 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-01-28 8:00 Andrej Picej
2022-01-31 9:23 ` Sascha Hauer
2022-01-31 10:23 ` Andrej Picej [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=813ef0f4-d6d4-1ca4-0729-09db490b811e@norik.com \
--to=andrej.picej@norik.com \
--cc=barebox@lists.infradead.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox