[NaCl SDK] nacl_io: Fix real_node when dealing with a windows terminal

[NaCl SDK] nacl_io: Fix real_node when dealing with a windows terminal

This fixes stat, read and write methods when dealing
with windows tty file descriptors. We were not catching
this is our tests because we run with stdout/stderr
that are not ttys.

Also, fail early and hard during nacl_io_init and ps_main
if anything goes wrong setting things up.

Committed: https://crrev.com/f08c3c1df1c2eed13054d6e6d1ad54e50e2919b1
Cr-Commit-Position: refs/heads/master@{#299005}

[Sam Clegg, 2014-10-08 01:07:29]

sbc@chromium.org changed reviewers:
+ binji@chromium.org

[binji, 2014-10-08 01:26:14]

lgtm

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
File native_client_sdk/src/libraries/nacl_io/kernel_proxy.cc (right):

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
native_client_sdk/src/libraries/nacl_io/kernel_proxy.cc:109: LOG_ERROR("failed
to /dev/stdin: %s", strerror(errno));
failed to open

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
native_client_sdk/src/libraries/nacl_io/kernel_proxy.cc:138: LOG_ERROR("error
initializeing streamfs: %s", strerror(result));
sp: initializing

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
native_client_sdk/src/libraries/nacl_io/kernel_proxy.cc:1212: LOG_TRACE("Invalid
timeout: tv_sec=%lld tv_usec=%ld.",
is this correct for all platforms?

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
File native_client_sdk/src/libraries/nacl_io/passthroughfs/real_node.cc (right):

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
native_client_sdk/src/libraries/nacl_io/passthroughfs/real_node.cc:40:
LOG_ERROR("_real_lseek failed: %s\n", strerror(err));
these could get noisy. Though I guess read/write/lseek don't typically fail.

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
native_client_sdk/src/libraries/nacl_io/passthroughfs/real_node.cc:75:
LOG_ERROR("_real_write failed: %d\n", err);
strerror

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
native_client_sdk/src/libraries/nacl_io/passthroughfs/real_node.cc:107: if
(S_ISCHR(stat->st_mode) && (stat->st_mode & S_IRWXU) == 0)
Maybe put this in a Windows-only ifdef?

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
File native_client_sdk/src/libraries/ppapi_simple/ps_main.cc (right):

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
native_client_sdk/src/libraries/ppapi_simple/ps_main.cc:46: if (rtn != 0)
Explicit ppapi_simple failure message here?

[Sam Clegg, 2014-10-08 02:01:02]

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
File native_client_sdk/src/libraries/nacl_io/kernel_proxy.cc (right):

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
native_client_sdk/src/libraries/nacl_io/kernel_proxy.cc:109: LOG_ERROR("failed
to /dev/stdin: %s", strerror(errno));
On 2014/10/08 01:26:13, binji wrote:
> failed to open

Done.

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
native_client_sdk/src/libraries/nacl_io/kernel_proxy.cc:138: LOG_ERROR("error
initializeing streamfs: %s", strerror(result));
On 2014/10/08 01:26:12, binji wrote:
> sp: initializing

Done.

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
native_client_sdk/src/libraries/nacl_io/kernel_proxy.cc:1212: LOG_TRACE("Invalid
timeout: tv_sec=%lld tv_usec=%ld.",
On 2014/10/08 01:26:13, binji wrote:
> is this correct for all platforms?

Done.

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
File native_client_sdk/src/libraries/nacl_io/passthroughfs/real_node.cc (right):

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
native_client_sdk/src/libraries/nacl_io/passthroughfs/real_node.cc:40:
LOG_ERROR("_real_lseek failed: %s\n", strerror(err));
On 2014/10/08 01:26:13, binji wrote:
> these could get noisy. Though I guess read/write/lseek don't typically fail.

Changed to LOG_WARN which are not compiled in by default ever for debug builds.

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
native_client_sdk/src/libraries/nacl_io/passthroughfs/real_node.cc:75:
LOG_ERROR("_real_write failed: %d\n", err);
On 2014/10/08 01:26:13, binji wrote:
> strerror

Done.

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
native_client_sdk/src/libraries/nacl_io/passthroughfs/real_node.cc:107: if
(S_ISCHR(stat->st_mode) && (stat->st_mode & S_IRWXU) == 0)
On 2014/10/08 01:26:13, binji wrote:
> Maybe put this in a Windows-only ifdef?

No, its the windows version of sel_ldr that is returning this info, so we can't
distinguish it at compile time or at run time :(

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
File native_client_sdk/src/libraries/ppapi_simple/ps_main.cc (right):

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
native_client_sdk/src/libraries/ppapi_simple/ps_main.cc:46: if (rtn != 0)
On 2014/10/08 01:26:13, binji wrote:
> Explicit ppapi_simple failure message here?

Using what mechanism.  If nacl_io init fails is there any point in writing to
stdout at this point?

[binji, 2014-10-08 18:22:25]

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
File native_client_sdk/src/libraries/nacl_io/passthroughfs/real_node.cc (right):

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
native_client_sdk/src/libraries/nacl_io/passthroughfs/real_node.cc:107: if
(S_ISCHR(stat->st_mode) && (stat->st_mode & S_IRWXU) == 0)
On 2014/10/08 02:01:02, Sam Clegg wrote:
> On 2014/10/08 01:26:13, binji wrote:
> > Maybe put this in a Windows-only ifdef?
> 
> No, its the windows version of sel_ldr that is returning this info, so we
can't
> distinguish it at compile time or at run time :(

Oh, I see.

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
File native_client_sdk/src/libraries/ppapi_simple/ps_main.cc (right):

https://codereview.chromium.org/635113002/diff/20001/native_client_sdk/src/li...
native_client_sdk/src/libraries/ppapi_simple/ps_main.cc:46: if (rtn != 0)
On 2014/10/08 02:01:02, Sam Clegg wrote:
> On 2014/10/08 01:26:13, binji wrote:
> > Explicit ppapi_simple failure message here?
> 
> Using what mechanism.  If nacl_io init fails is there any point in writing to
> stdout at this point?

I was thinking a direct write to stderr. Does nacl_io_init always print an error
if it fails to initialize? Even if so, this could be a useful fallback so the
user can see quickly what is failing -- nacl_io_init.

[Sam Clegg, 2014-10-09 22:28:51]

The CQ bit was checked by sbc@chromium.org

[commit-bot: I haz the power, 2014-10-09 22:30:04]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/635113002/60001

[commit-bot: I haz the power, 2014-10-09 23:12:09]

Committed patchset #4 (id:60001)

[commit-bot: I haz the power, 2014-10-09 23:13:41]

Patchset 4 (id:??) landed as
https://crrev.com/f08c3c1df1c2eed13054d6e6d1ad54e50e2919b1
Cr-Commit-Position: refs/heads/master@{#299005}