kernel/linux.git/drivers/net/ipa/ipa_table.c, branch v6.6.132

net: ipa: only reset hashed tables when supported

2023-07-26T03:34:20+00:00

Last year, the code that manages GSI channel transactions switched from using spinlock-protected linked lists to using indexes into the ring buffer used for a channel. Recently, Google reported seeing transaction reference count underflows occasionally during shutdown. Doug Anderson found a way to reproduce the issue reliably, and bisected the issue to the commit that eliminated the linked lists and the lock. The root cause was ultimately determined to be related to unused transactions being committed as part of the modem shutdown cleanup activity. Unused transactions are not normally expected (except in error cases). The modem uses some ranges of IPA-resident memory, and whenever it shuts down we zero those ranges. In ipa_filter_reset_table() a transaction is allocated to zero modem filter table entries. If hashing is not supported, hashed table memory should not be zeroed. But currently nothing prevents that, and the result is an unused transaction. Something similar occurs when we zero routing table entries for the modem. By preventing any attempt to clear hashed tables when hashing is not supported, the reference count underflow is avoided in this case. Note that there likely remains an issue with properly freeing unused transactions (if they occur due to errors). This patch addresses only the underflows that Google originally reported. Cc: # 6.1.x Fixes: d338ae28d8a8 ("net: ipa: kill all other transaction lists") Tested-by: Douglas Anderson Signed-off-by: Alex Elder Link: https://lore.kernel.org/r/20230724224055.1688854-1-elder@linaro.org Signed-off-by: Jakub Kicinski

net: ipa: generalize register field functions

2023-02-10T08:06:32+00:00

Rename functions related to register fields so they don't appear to be IPA-specific, and move their definitions into "reg.h": ipa_reg_fmask() -> reg_fmask() ipa_reg_bit() -> reg_bit() ipa_reg_field_max() -> reg_field_max() ipa_reg_encode() -> reg_encode() ipa_reg_decode() -> reg_decode() Signed-off-by: Alex Elder Signed-off-by: David S. Miller

net: ipa: generalize register offset functions

2023-02-10T08:06:32+00:00

Rename ipa_reg_offset() to be reg_offset() and move its definition to "reg.h". Rename ipa_reg_n_offset() to be reg_n_offset() also. Signed-off-by: Alex Elder Signed-off-by: David S. Miller

net: ipa: start generalizing "ipa_reg"

2023-02-10T08:06:32+00:00

IPA register definitions have evolved with each new version. The changes required to support more than 32 endpoints in IPA v5.0 made it best to define a unified mechanism for defining registers and their fields. GSI register definitions, meanwhile, have remained fairly stable. And even as the total number of IPA endpoints goes beyond 32, the number of GSI channels on a given EE that underly endpoints still remains 32 or less. Despite that, GSI v3.0 (which is used with IPA v5.0) extends the number of channels (and events) it supports to be about 256, and as a result, many GSI register definitions must change significantly. To address this, we'll use the same "ipa_reg" mechanism to define the GSI registers. As a first step in generalizing the "ipa_reg" to also support GSI registers, isolate the definitions of the "ipa_reg" and "ipa_regs" structure types (and some supporting macros) into a new header file, and remove the "ipa_" and "IPA_" from symbol names. Separate the IPA register ID validity checking from the generic check that a register ID is in range. Aside from that, this is intended to have no functional effect on the code. Signed-off-by: Alex Elder Signed-off-by: David S. Miller

net: ipa: support zeroing new cache tables

2023-02-01T05:45:52+00:00

IPA v5.0+ separates the configuration of entries in the cached (previously "hashed") routing and filtering tables into distinct registers. Previously a single "filter and router" register updated entries in both tables at once; now the routing and filter table caches have separate registers that define their content. This patch updates the code that zeroes entries in the cached filter and router tables to support IPA versions including v5.0+. Signed-off-by: Alex Elder Reviewed-by: Leon Romanovsky Signed-off-by: Jakub Kicinski

net: ipa: update table cache flushing

2023-02-01T05:45:51+00:00

Update the code that causes filter and router table caches to be flushed so that it supports IPA versions 5.0+. It adds a comment in ipa_hardware_config_hashing() that explains that cacheing does not need to be enabled, just as before, because it's enabled by default. (For the record, the FILT_ROUT_CACHE_CFG register would have been used if we wanted to explicitly enable these.) Signed-off-by: Alex Elder Reviewed-by: Leon Romanovsky Signed-off-by: Jakub Kicinski

net: ipa: avoid a null pointer dereference

2022-11-18T11:46:18+00:00

Dan Carpenter reported that Smatch found an instance where a pointer which had previously been assumed could be null (as indicated by a null check) was later dereferenced without a similar check. In practice this doesn't lead to a problem because currently the pointers used are all non-null. Nevertheless this patch addresses the reported problem. In addition, I spotted another bug that arose in the same commit. When the command to initialize a routing table memory region was added, the number of entries computed for the non-hashed table was wrong (it ended up being a Boolean rather than the count intended). This bug is fixed here as well. Reported-by: Dan Carpenter Link: https://lore.kernel.org/kernel-janitors/Y3OOP9dXK6oEydkf@kili Tested-by: Caleb Connolly Fixes: 5cb76899fb47 ("net: ipa: reduce arguments to ipa_table_init_add()") Signed-off-by: Alex Elder Signed-off-by: David S. Miller

net: ipa: support more filtering endpoints

2022-11-04T10:16:53+00:00

Prior to IPA v5.0, there could be no more than 32 endpoints. A filter table begins with a bitmap indicating which endpoints have a filter defined. That bitmap is currently assumed to fit in a 32-bit value. Starting with IPA v5.0, more than 32 endpoints are supported, so it's conceivable that a TX endpoint has an ID that exceeds 32. Increase the size of the field representing endpoints that support filtering to 64 bits. Rename the bitmap field "filtered". Unlike other similar fields, we do not use an (arbitrarily long) Linux bitmap for this purpose. The reason is that if a filter table ever *did* need to support more than 64 TX endpoints, its format would change in ways we can't anticipate. Have ipa_endpoint_init() return a negative errno rather than a mask that indicates which endpoints support filtering, and have that function assign the "filtered" field directly. Signed-off-by: Alex Elder Signed-off-by: David S. Miller

net: ipa: use ipa_table_mem() in ipa_table_reset_add()

2022-11-04T10:16:53+00:00

Similar to the previous commit, pass flags rather than a memory region ID to ipa_table_reset_add(), and there use ipa_table_mem() to look up the memory region affected based on those flags. Currently all eight of these table memory regions are assumed to exist, because they all have canaries within them. Stop assuming that will always be the case, and in ipa_table_reset_add() allow these memory regions to be non-existent. Signed-off-by: Alex Elder Signed-off-by: David S. Miller

net: ipa: reduce arguments to ipa_table_init_add()

2022-11-04T10:16:53+00:00

Recently ipa_table_mem() was added as a way to look up one of 8 possible memory regions by indicating whether it was a filter or route table, hashed or not, and IPv6 or not. We can simplify the interface to ipa_table_init_add() by passing two flags to it instead of the opcode and both hashed and non-hashed memory region IDs. The "filter" and "ipv6" flags are sufficient to determine the opcode to use, and with ipa_table_mem() can look up the correct memory region as well. It's possible to not have hashed tables, but we already verify the number of entries in a filter or routing table is nonzero. Stop assuming a hashed table entry exists in ipa_table_init_add(). Signed-off-by: Alex Elder Signed-off-by: David S. Miller