[APS IR] Minfor fix - use GetAttrKey in get_keystr to match with flat…

… args path in unflatten (pytorch#161453)

Summary: While passing path info to [_check_input_constraints_for_graph](https://www.internalfb.com/code/fbsource/[6b5b2dc35902a26ce265e3c0ae5189a3faba1d38]/fbcode/caffe2/torch/export/unflatten.py?lines=594), GetAttrKey is used to specify path str. To match with that get_keystr should also use GetAttrKey.

Test Plan:
Existing tests

```
buck run mode/opt caffe2/test:test_export -- -r unflatten
```

```
Ran 413 tests in 204.533s

OK (skipped=1, expected failures=13)
```

Rollback Plan:

Differential Revision: D80984083

Pull Request resolved: pytorch#161453
Approved by: https://github.com/tugsbayasgalan

[BE] fix compute_global_tensor_shape test (pytorch#161441)

Fixes pytorch#161154

**Test**
`pytest  test/distributed/tensor/test_utils.py -s -k test_compute_global_tensor_shape_1D`

Pull Request resolved: pytorch#161441
Approved by: https://github.com/kwen2501

Revert "[dynamo, nested graph breaks] add nested graph break tests (p…

…ytorch#144516)"

This reverts commit 9a756c2.

Reverted pytorch#144516 on behalf of https://github.com/atalman due to failing internal tests ([comment](pytorch#144516 (comment)))

[ROCm][CI] restore test_flex_attention tests (pytorch#161519)

Reverts pytorch#161450 and targets specific subtests to skip on MI200.

Pull Request resolved: pytorch#161519
Approved by: https://github.com/jeffdaily

Co-authored-by: Jeff Daily <[email protected]>

[reland] [dynamo] Refactor convert_frame.compile_frame to be self con…

…tained function. [5/n] (pytorch#161514)

Summary:
convert_frame.compile_frame used to take a callback transform function which will capture the frame object it has, but the frame information is not passed directly into compile_frame function.

This PR changes the signature of compile_frame so that frame information is directly passed in the function without taking a callback. This makes it easier to build fullgraph capture API on top of compile_frame.

Test Plan:
CI

Rollback Plan:

Differential Revision: D81041296

Pull Request resolved: pytorch#161514
Approved by: https://github.com/tugsbayasgalan

Revert "[dynamo, nested graph breaks] prevent excessive recompilations (

pytorch#159786)"

This reverts commit 67d31f6.

Reverted pytorch#159786 on behalf of https://github.com/atalman due to failing internal tests ([comment](pytorch#159786 (comment)))

[dynamo] auto lift unbacked symbol in tensor's storage_offset (pytorc…

…h#161199)

```python
import torch

torch._dynamo.config.capture_scalar_outputs = True

class M(torch.nn.Module):
    def forward(self, idx, x):
        u0 = idx.item()
        x0 = x.select(0, u0)
        def fn():
            return x0.sin()
        return torch.cond(x0.sum() > 0, fn, fn)

m = M()
out = torch.compile(m, fullgraph=True)(torch.tensor(0, dtype=torch.int64, device="cuda"), torch.randn(3, 3, device="cuda"))
print(out)

```

Before the PR, we didn't track the storage_offset symbol of a tensor. After pytorch#157605, we create an unbacked_symint for stroage_offset for the result of select. So when we try to lift the free basic symbols of x0  during speculating fn, we found a free symbol that's not bound to a proxy.

This PR tracks the symbols of storage_offset and associated it with a proxy using torch.ops.aten.storage_offest.

Pull Request resolved: pytorch#161199
Approved by: https://github.com/zou3519
ghstack dependencies: pytorch#161198

Revert "Increase timeout value when pushing to ghcr.io (pytorch#161444)"

This reverts commit b9e9e92.

Reverted pytorch#161444 on behalf of https://github.com/huydhn due to Reland this to generate a different has value for the benchmark Docker image ([comment](pytorch#161444 (comment)))

Revert "Ensure large tensor int32 -> int64 indexing is enabled (pytor…

…ch#157767)"

This reverts commit fc69c2b.

Reverted pytorch#157767 on behalf of https://github.com/atalman due to internal failure, sorry will revert ([comment](pytorch#157767 (comment)))

[Inductor][Triton] Fix SCALING_ROWWISE misclassification for scalar s…

…cales (pytorch#160450)

Summary: In `tuned_scaled_mm()`, we unsqeeuze any scalar scale from [] -> [1, 1]. Later, when we are determining how to set the `SCALING_ROWWISE` kernel attribute, we check whether the scale has 2 dimensions. However, since we previously unsqueezed any scalar scales, this will always evaluate to True.

Test Plan:
Run the following tests in test/inductor/test_fp8.py:
test_tensorwise_scaling_tma_template
test_rowwise_scaling_tma_template

Rollback Plan:

Differential Revision: D80108117

Pull Request resolved: pytorch#160450
Approved by: https://github.com/eellison