threads / patch / 44737

patchmingw: consider that UNICODE_STRING::Length counts bytes

Subject: [PATCH] mingw: consider that UNICODE_STRING::Length counts bytes

## tl;dr

5 messages between Dec 19, 2016 and Dec 20, 2016. Diffs are folded; open one to read it.

replies: 4people: 3as markdown or json

Max Kirillov· Dec 19, 2016, 21:32 UTC · lore

UNICODE_STRING::Length field means size of buffer in bytes[1], despite of buffer itself being array of wchar_t. Because of that terminating zero is placed twice as far. Fix it.

[1] https://msdn.microsoft.com/en-us/library/windows/desktop/aa380518.aspx
Signed-off-by: Max Kirillov <max@max630.net>
---
Access outside of buffer was very unlikely (for that user needed to redirect
standard fd to a file with path longer than ~250 symbols), it still did not
seem to do any harm, and otherwise it did not break because only substring is
checked, but it was still incorrect.
 compat/winansi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Show changes to compat/winansi.c +1 −1
diff --git a/compat/winansi.c b/compat/winansi.c
index 3be60ce..6b4f736 100644
--- a/compat/winansi.c
+++ b/compat/winansi.c
@@ -553,7 +553,7 @@ static void detect_msys_tty(int fd)
 			buffer, sizeof(buffer) - 2, &result)))
 		return;
 	name = nameinfo->Name.Buffer;
-	name[nameinfo->Name.Length] = 0;
+	name[nameinfo->Name.Length / sizeof(*name)] = 0;
 
 	/* check if this could be a MSYS2 pty pipe ('msys-XXXX-ptyN-XX') */
 	if (!wcsstr(name, L"msys-") || !wcsstr(name, L"-pty"))
-- 
2.3.4.2801.g3d0809b
Junio C Hamano· Dec 19, 2016, 21:57 UTC · re: Max Kirillov · lore

Re: [PATCH] mingw: consider that UNICODE_STRING::Length counts bytes

Max Kirillov <max@max630.net> writes:
Show 8 quoted lines
> UNICODE_STRING::Length field means size of buffer in bytes[1], despite of buffer
> itself being array of wchar_t. Because of that terminating zero is placed twice
> as far. Fix it.
>
> [1] https://msdn.microsoft.com/en-us/library/windows/desktop/aa380518.aspx
>
> Signed-off-by: Max Kirillov <max@max630.net>
> ---

Max, I see this is a resend from a few days ago. I suspect that we are in a slow season, so there may be longer delay until we see responses to a patch.

I will wait until taking any patch to files specific to MS Windows platform (compat/win*, compat/mingw*, and compat/vcbuild*) without first seeing it reviewed and acked by either of the two Johannes's (well, there might be other people in addition to those two, whose Acked-by/Reviewed-by I should be trusting; if that is the case, it only shows how unqualified I would be as the first contact to be on the To: line of such a patch).

I do not mind "see the patch, ping Johanneses as needed, wait and then apply it to my tree" flow, but I also wonder if the process can be made more efficient. Dscho (one of the Johanneses) gets many patches specific to Windows port via the Git-for-Windows project and then "upstreams" by forwarding with his sign-off in my direction, and I do not mind that flow, either. Whichever one is the most efficient for all three parties involved is fine by me, but if one is preferred over the other, perhaps I should write it down in the "notes from the maintainer" or somewhere?

Dscho, J6t, what do you think?
Thanks.
Show 20 quoted lines
> Access outside of buffer was very unlikely (for that user needed to redirect
> standard fd to a file with path longer than ~250 symbols), it still did not
> seem to do any harm, and otherwise it did not break because only substring is
> checked, but it was still incorrect.
>  compat/winansi.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/compat/winansi.c b/compat/winansi.c
> index 3be60ce..6b4f736 100644
> --- a/compat/winansi.c
> +++ b/compat/winansi.c
> @@ -553,7 +553,7 @@ static void detect_msys_tty(int fd)
>  			buffer, sizeof(buffer) - 2, &result)))
>  		return;
>  	name = nameinfo->Name.Buffer;
> -	name[nameinfo->Name.Length] = 0;
> +	name[nameinfo->Name.Length / sizeof(*name)] = 0;
>  
>  	/* check if this could be a MSYS2 pty pipe ('msys-XXXX-ptyN-XX') */
>  	if (!wcsstr(name, L"msys-") || !wcsstr(name, L"-pty"))
Max Kirillov· Dec 20, 2016, 05:21 UTC · re: Junio C Hamano · lore

Re: [PATCH] mingw: consider that UNICODE_STRING::Length counts bytes

On Mon, Dec 19, 2016 at 01:57:29PM -0800, Junio C Hamano wrote:
> Max, I see this is a resend from a few days ago

Sorry about resend. For some reason I don't get the list copy (might be some clever duplicate elimination in my forwardings), and marc.info seems to be slow to update, so I had no indication the first message got into list. Now I see they are there in the other archive.

PS: probably http://vger.kernel.org/vger-lists.html#git
should be updated because there is no archive at gmane
anymore.
-- 
Max
Johannes Schindelin· Dec 20, 2016, 15:16 UTC · re: Max Kirillov · lore

Re: [PATCH] mingw: consider that UNICODE_STRING::Length counts bytes

Hi Max,
On Mon, 19 Dec 2016, Max Kirillov wrote:
> UNICODE_STRING::Length field means size of buffer in bytes[1], despite
> of buffer itself being array of wchar_t. Because of that terminating
> zero is placed twice as far. Fix it.
This commit message needs to be wrapped at <= 76 columns per row.
Show 8 quoted lines
> [1] https://msdn.microsoft.com/en-us/library/windows/desktop/aa380518.aspx
> 
> Signed-off-by: Max Kirillov <max@max630.net>
> ---
> Access outside of buffer was very unlikely (for that user needed to redirect
> standard fd to a file with path longer than ~250 symbols), it still did not
> seem to do any harm, and otherwise it did not break because only substring is
> checked, but it was still incorrect.

Very good, thank you for fixing my mistake! I verified locally that it does exactly the right thing with your patch.

ACK, Dscho

Junio C Hamano· Dec 20, 2016, 17:07 UTC · re: Johannes Schindelin · lore

Re: [PATCH] mingw: consider that UNICODE_STRING::Length counts bytes

Johannes Schindelin <Johannes.Schindelin@gmx.de> writes:
Show 12 quoted lines
> Hi Max,
>
> On Mon, 19 Dec 2016, Max Kirillov wrote:
>
>> UNICODE_STRING::Length field means size of buffer in bytes[1], despite
>> of buffer itself being array of wchar_t. Because of that terminating
>> zero is placed twice as far. Fix it.
>
> This commit message needs to be wrapped at <= 76 columns per row.
> ...
> Very good, thank you for fixing my mistake! I verified locally that it
> does exactly the right thing with your patch.
Thanks, both.  I'll queue this like so.
-- >8 --
From: Max Kirillov <max@max630.net>
Date: Mon, 19 Dec 2016 23:32:00 +0200
Subject: [PATCH] mingw: consider that UNICODE_STRING::Length counts bytes

UNICODE_STRING::Length field means size of buffer in bytes[1], despite of buffer itself being array of wchar_t. Because of that terminating zero is placed twice as far. Fix it.

[1] https://msdn.microsoft.com/en-us/library/windows/desktop/aa380518.aspx
Signed-off-by: Max Kirillov <max@max630.net>
Acked-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 compat/winansi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Show changes to compat/winansi.c +1 −1
diff --git a/compat/winansi.c b/compat/winansi.c
index 3be60ce1c6..6b4f736fdc 100644
--- a/compat/winansi.c
+++ b/compat/winansi.c
@@ -553,7 +553,7 @@ static void detect_msys_tty(int fd)
 			buffer, sizeof(buffer) - 2, &result)))
 		return;
 	name = nameinfo->Name.Buffer;
-	name[nameinfo->Name.Length] = 0;
+	name[nameinfo->Name.Length / sizeof(*name)] = 0;
 
 	/* check if this could be a MSYS2 pty pipe ('msys-XXXX-ptyN-XX') */
 	if (!wcsstr(name, L"msys-") || !wcsstr(name, L"-pty"))

base-commit: f7f90e0f4f58d493242078d17c0eba41dd3f1f79
-- 
2.11.0-416-g1351c11cce

← back to recent threads