Page MenuHomeFreeBSD

jhb (John Baldwin)
User

Projects (8)

User Details

User Since
Mar 11 2014, 8:46 PM (271 w, 4 d)

Recent Activity

Yesterday

jhb added a comment to D20403: devel/gdb: Update port to GDB 8.3..

All of the removed patches were issues that were fixed differently upstream. Having those patches just to fix warnings did make the process of doing the upgrade take a bit longer however.

Sat, May 25, 1:20 AM
jhb added a reviewer for D20403: devel/gdb: Update port to GDB 8.3.: luca.pizzamiglio_gmail.com.
Sat, May 25, 1:17 AM
jhb created D20403: devel/gdb: Update port to GDB 8.3..
Sat, May 25, 1:17 AM

Fri, May 24

jhb committed rS348254: Restructure mbuf send tags to provide stronger guarantees..
Restructure mbuf send tags to provide stronger guarantees.
Fri, May 24, 11:31 PM
jhb closed D20117: Restructure mbuf send tags to provide stronger guarantees..
Fri, May 24, 11:30 PM
jhb accepted D20401: Disable intr_storm_threshold mechanism by default.

My only suggestion would perhaps be to add some text along these lines to the commit log:

Fri, May 24, 10:29 PM
jhb committed rS348253: Add initial support for 'qSupported' to the debug server..
Add initial support for 'qSupported' to the debug server.
Fri, May 24, 10:12 PM
jhb closed D20308: Add initial support for qSupported..
Fri, May 24, 10:11 PM
jhb accepted D19298: AESNI version of CCM+CBC-MAC.
Fri, May 24, 9:19 PM
jhb added a comment to D19506: Fix bug in bhyve by ignoring writes to errata MSRs in AMD cpus.

@jhb, reopening a commited phab review messes with things, as now when you make a second commit the default top patch in this review well be the one you commit, and one has to go digging in the review history to find the prior patch that was in the first commit. It is fine to reopen a review for a revert, but I think one should start a new review for a new change, which is what this is.

Fri, May 24, 2:53 PM · bhyve
jhb committed rS348213: MFC 340486: Consolidate on a single set of constants for SCMD fields..
MFC 340486: Consolidate on a single set of constants for SCMD fields.
Fri, May 24, 12:44 AM
jhb committed rS348212: Add support for writing to guest memory in the debug server..
Add support for writing to guest memory in the debug server.
Fri, May 24, 12:34 AM
jhb closed D20307: Add support for writing to guest memory in the debug server..
Fri, May 24, 12:34 AM
jhb added a comment to D19841: bhyve NVMe emulation locking.

So I think splitting the queue init out of reset is probably a good change to do on its own as a separate commit. I think it will also make the locking part of the diff easier to see.

Fri, May 24, 12:20 AM

Thu, May 23

jhb added a comment to D19506: Fix bug in bhyve by ignoring writes to errata MSRs in AMD cpus.

José, would you be able to test the updated version that permits read as well as write and verify it still works for your test case? It would be sufficient to just update your local patch to enable reading, no need to rebase on top of what I've committed so far.

Thu, May 23, 11:59 PM · bhyve
jhb updated the diff for D19506: Fix bug in bhyve by ignoring writes to errata MSRs in AMD cpus.
  • Support read and write.
Thu, May 23, 11:56 PM · bhyve
jhb added a comment to D19506: Fix bug in bhyve by ignoring writes to errata MSRs in AMD cpus.

FreeBSD at least expects to be able to read from this MSR as well as write it, though FreeBSD is careful to only do this on bare metal.

Thu, May 23, 11:54 PM · bhyve
jhb reopened D19506: Fix bug in bhyve by ignoring writes to errata MSRs in AMD cpus.
Thu, May 23, 11:40 PM · bhyve
jhb added a comment to D19506: Fix bug in bhyve by ignoring writes to errata MSRs in AMD cpus.

I've chosen to name the constant MSR_LS_CFG since that seems more consistent with how FreeBSD has chosen constants for AMD MSRs. I've also put MSR_LS_CFG next to MSR_IC_CFG in an earlier section in xmsr.c.

Thu, May 23, 11:39 PM · bhyve
jhb committed rS348210: Add a constant for the LS config MSR on AMD CPUs..
Add a constant for the LS config MSR on AMD CPUs.
Thu, May 23, 11:37 PM
jhb closed D19506: Fix bug in bhyve by ignoring writes to errata MSRs in AMD cpus.
Thu, May 23, 11:37 PM · bhyve
jhb closed D20344: Add deprecation warnings for weaker algorithms to geli(4)..
Thu, May 23, 10:32 PM
jhb committed rS348206: Add deprecation warnings for weaker algorithms to geli(4)..
Add deprecation warnings for weaker algorithms to geli(4).
Thu, May 23, 10:32 PM
jhb committed rS348205: Add deprecation warnings for IPsec algorithms deprecated in RFC 8221..
Add deprecation warnings for IPsec algorithms deprecated in RFC 8221.
Thu, May 23, 10:07 PM
jhb closed D20340: Add deprecation warnings for IPsec algorithms deprecated in RFC 8221..
Thu, May 23, 10:07 PM
jhb committed rS348203: MFC 346617: Test the AES-CCM test vectors from the NIST Known Answer Tests..
MFC 346617: Test the AES-CCM test vectors from the NIST Known Answer Tests.
Thu, May 23, 9:58 PM
jhb committed rS348202: MFC 346649: Don't panic for empty CCM requests..
MFC 346649: Don't panic for empty CCM requests.
Thu, May 23, 9:52 PM
jhb committed rS348200: MFC 346648: Fix requests for "plain" SHA digests of an empty buffer..
MFC 346648: Fix requests for "plain" SHA digests of an empty buffer.
Thu, May 23, 9:05 PM
jhb added inline comments to D20344: Add deprecation warnings for weaker algorithms to geli(4)..
Thu, May 23, 8:56 PM
jhb updated the diff for D20340: Add deprecation warnings for IPsec algorithms deprecated in RFC 8221..
  • Add individual warn variables.
Thu, May 23, 8:41 PM
jhb committed rS348193: MFC 346616: Run the plain SHA digest tests from NIST..
MFC 346616: Run the plain SHA digest tests from NIST.
Thu, May 23, 7:20 PM
jhb committed rS348192: MFC 346615: Use more descriptive algorithm names in skip messages..
MFC 346615: Use more descriptive algorithm names in skip messages.
Thu, May 23, 7:19 PM
jhb committed rS348191: MFC 346614: Skip tests with missing test vectors instead of failing..
MFC 346614: Skip tests with missing test vectors instead of failing.
Thu, May 23, 7:18 PM
jhb committed rS348190: MFC 346421: Test SHA2-224-HMAC now that OCF supports it..
MFC 346421: Test SHA2-224-HMAC now that OCF supports it.
Thu, May 23, 7:14 PM
jhb committed rS348189: MFC 346419: Sync cryptographic algorithm constants with current cryptodev.h..
MFC 346419: Sync cryptographic algorithm constants with current cryptodev.h.
Thu, May 23, 7:13 PM
jhb added inline comments to D19905: bhyve: Add EUI64 to NVMe device.
Thu, May 23, 5:17 PM
jhb accepted D19904: bhyve: Add Integrated Endpoint to PCIe Capability.

We should fix hostbridge devices to be endpoints as a followup, and we should sprinkle pci_emul_add_pciecap in more places (ahci and virtio).

Thu, May 23, 4:50 PM
jhb added a comment to D20365: simplify bhyve vlapic ESR logic.

I think it would actually be a smaller and simpler diff to add a 'firing_error' bool to vlapic_fire_lvt() that is normally false except in vlapic_set_error() and then have the vlapic_fire_lvt() pass that down rather through the various layers. OTOH, I'm not really sure that is a lot better than esr_firing and just ignoring the field during suspend and restore?

Thu, May 23, 4:32 PM · bhyve

Wed, May 22

jhb added a comment to D17512: Fix dpcpu and vnet panics with complex types at the end of the section.

I still don't understand why i386 is special in this regard. Surely i386 isn't the only architecture to ever use this type of relocation?

Wed, May 22, 10:56 PM
jhb added inline comments to D20340: Add deprecation warnings for IPsec algorithms deprecated in RFC 8221..
Wed, May 22, 12:18 AM
jhb added a comment to D20343: Add warnings for Kerberos GSS algorithms deprecated in RFCs 6649 and 8429..

From my code reading these init routines should be called rarely (during session setup as opposed to per-packet). If someone is able to test this and finds it is in fact spammy I'd be happy to rate limit it.

Wed, May 22, 12:11 AM

Tue, May 21

jhb updated the diff for D20117: Restructure mbuf send tags to provide stronger guarantees..
  • Rebase after ip6_fragment commit.
Tue, May 21, 11:10 PM
jhb added a comment to D20340: Add deprecation warnings for IPsec algorithms deprecated in RFC 8221..

I wasn't planning on doing ipcomp yet, but might propose a diff for that next.

Tue, May 21, 10:46 PM
jhb added inline comments to D20340: Add deprecation warnings for IPsec algorithms deprecated in RFC 8221..
Tue, May 21, 10:26 PM
jhb created D20344: Add deprecation warnings for weaker algorithms to geli(4)..
Tue, May 21, 10:17 PM
jhb created D20343: Add warnings for Kerberos GSS algorithms deprecated in RFCs 6649 and 8429..
Tue, May 21, 10:13 PM
jhb updated the diff for D20340: Add deprecation warnings for IPsec algorithms deprecated in RFC 8221..
  • Drop extra blank line.
Tue, May 21, 10:12 PM
jhb added inline comments to D20340: Add deprecation warnings for IPsec algorithms deprecated in RFC 8221..
Tue, May 21, 10:10 PM
jhb created D20340: Add deprecation warnings for IPsec algorithms deprecated in RFC 8221..
Tue, May 21, 9:40 PM
jhb added a comment to D20327: Don't reset memory attributes when mapping physical addresses for ACPI..
In D20327#438513, @kib wrote:
In D20327#438503, @jhb wrote:

What happens on the laptops today is that it calls AcpiOsMapMemory for the MCFG region which uses pmap_mapbios and switches the mapping in the direct map from UC to WB and flushes the cache which then hangs. What patch is changing is making AcpiOsMapMemory just leave the direct map as-is and if the requested address has been marked UC prior to the use of AcpiOsMapMemory, it just leaves it alone instead of changing it to WB. I am actually inclined to do this for all the pmap_mapbios calls by having them always honor existing attributes which was my first question above. A further question at that point is if we'd rather just use the new name always since 'mapbios' is kind of a crummy name. Originally 'mapbios' meant 'map a table provided by the BIOS', but given that AML is free to map whatever random thing it chooses, 'mapphys' is probably a better name (and more generic for platforms that don't have BIOS).

Yes, I both do not like that pmap_mapbios() does not change the attributes for random requests ( I believe that GPU apertures are mapped by pmap_mapbios()), and that it could be done more explicit, snce we actually care about MFCG. Let me provide more or less accurate sketch of what I mean. The code duplication can be optimized out, I am more about the approach.

Tue, May 21, 9:33 PM
jhb added a comment to D20307: Add support for writing to guest memory in the debug server..

I did consider merging the read and write mem functions, but I was worried it might be a bit messy. I might still try it as a followup and see how it looks.

Tue, May 21, 9:16 PM
jhb added a comment to D20327: Don't reset memory attributes when mapping physical addresses for ACPI..

What happens on the laptops today is that it calls AcpiOsMapMemory for the MCFG region which uses pmap_mapbios and switches the mapping in the direct map from UC to WB and flushes the cache which then hangs. What patch is changing is making AcpiOsMapMemory just leave the direct map as-is and if the requested address has been marked UC prior to the use of AcpiOsMapMemory, it just leaves it alone instead of changing it to WB. I am actually inclined to do this for all the pmap_mapbios calls by having them always honor existing attributes which was my first question above. A further question at that point is if we'd rather just use the new name always since 'mapbios' is kind of a crummy name. Originally 'mapbios' meant 'map a table provided by the BIOS', but given that AML is free to map whatever random thing it chooses, 'mapphys' is probably a better name (and more generic for platforms that don't have BIOS).

Tue, May 21, 6:44 PM
jhb added a comment to D20327: Don't reset memory attributes when mapping physical addresses for ACPI..
In D20327#438343, @kib wrote:

Is it too early for that problematic calls to pmap_mapdev_attr() to notice that the request is really to MCFG window and redirect the call to pmap_mapdev_pciecfg() ?

Tue, May 21, 4:03 PM

Mon, May 20

jhb added a comment to D20327: Don't reset memory attributes when mapping physical addresses for ACPI..

This isn't a final review yet in the sense that it needs more work before it can be committed. I wanted to get some thoughts though on how best to do this. Specifically, we could perhaps repurpose pmap_mapbios() and give it the semantics of pmap_mapphys() either by changing the implementation of pmap_mapbios() to have the semantics of the new thing (don't force WB, just map whatever is there). Alternatively, we could retire pmap_mapbios() and always use the new name if we think the new name is a better one in that case.

Mon, May 20, 10:41 PM
jhb created D20327: Don't reset memory attributes when mapping physical addresses for ACPI..
Mon, May 20, 10:34 PM
jhb added a comment to D19247: Merge hold_count into wire_count..

The cxgbe changes are ok.

Mon, May 20, 10:15 PM
jhb added a comment to D18291: Align mips crt .init/.fini sections to 4 bytes instead of 16.

Hmm, never mind then. CheriBSD is using .balign but I assumed that meant some power-of-two thing (so .balign 4 == align to 16). Still didn't answer my question of if you ever tested this. :-P

Mon, May 20, 10:07 PM
jhb added a comment to D18291: Align mips crt .init/.fini sections to 4 bytes instead of 16.

(FWIW, the last time I tried I couldn't get clang 8 to even build a 32-bit mips, I didn't try mips64 though)

Mon, May 20, 9:21 PM
jhb added a comment to D18291: Align mips crt .init/.fini sections to 4 bytes instead of 16.

Did you test this? In CheriBSD we are definitely using the equivalent of .align 16. MIPS instructions are only 4-bytes, so instruction alignment as described in your latest comment would seem to imply that '.align 4' was redundant?

Mon, May 20, 9:20 PM
jhb added a comment to D20326: Expose the buildid hash through sysctl.

Just to be clear, this doesn't compile yet on !amd64? Also, does uname -a still work for a kernel built without build-id?

Mon, May 20, 9:07 PM
jhb committed rP502170: New port: devel/git-subrepo: Git command for managing subrepositories.
New port: devel/git-subrepo: Git command for managing subrepositories
Mon, May 20, 9:04 PM
jhb closed D20189: [new port] devel/git-subrepo, a Git command for managing subrepositories..
Mon, May 20, 9:04 PM
jhb added a comment to D20189: [new port] devel/git-subrepo, a Git command for managing subrepositories..

bdrewery looked this over and approved it in person during a break in between talks at BSDCan

Mon, May 20, 6:29 PM

Sat, May 18

jhb closed D20296: bhyve should expose MD_CLEAR capability.
Sat, May 18, 9:21 PM · bhyve
jhb committed rS347964: Expose the MD_CLEAR capability used by Intel MDS mitigations to guests..
Expose the MD_CLEAR capability used by Intel MDS mitigations to guests.
Sat, May 18, 9:20 PM
jhb added a comment to D20296: bhyve should expose MD_CLEAR capability.

I haven't personally made a conscious decision one way or another FWIW. If we should be exposing some of the MSRs and flag bits we can add that. I just wasn't sure which ones were relevant for guests (e.g. was L1TF only really an issue for hypervisors rather than something a guest or bare-metal kernel needs)

Sat, May 18, 9:14 PM · bhyve
jhb added a comment to D20296: bhyve should expose MD_CLEAR capability.

I'm going to commit this in a bit, but I wanted to get @kib's opinion on whether there are other bits from other mitigations that we should be exposing to guests?

Sat, May 18, 8:45 PM · bhyve
jhb updated subscribers of D20296: bhyve should expose MD_CLEAR capability.
Sat, May 18, 8:43 PM · bhyve
jhb accepted D20311: Change ed(4), ep(4), and fxp(4) examples to em(4)..
Sat, May 18, 8:32 PM
jhb added a child revision for D20307: Add support for writing to guest memory in the debug server.: D20308: Add initial support for qSupported..
Sat, May 18, 6:36 PM
jhb added a parent revision for D20308: Add initial support for qSupported.: D20307: Add support for writing to guest memory in the debug server..
Sat, May 18, 6:36 PM
jhb added a child revision for D20308: Add initial support for qSupported.: D20309: Support software breakpoints in the debug server..
Sat, May 18, 6:33 PM
jhb added a parent revision for D20309: Support software breakpoints in the debug server.: D20308: Add initial support for qSupported..
Sat, May 18, 6:33 PM
jhb removed a parent revision for D20308: Add initial support for qSupported.: D20309: Support software breakpoints in the debug server..
Sat, May 18, 6:33 PM
jhb removed a child revision for D20309: Support software breakpoints in the debug server.: D20308: Add initial support for qSupported..
Sat, May 18, 6:33 PM
jhb added a parent revision for D20308: Add initial support for qSupported.: D20309: Support software breakpoints in the debug server..
Sat, May 18, 6:32 PM
jhb added a child revision for D20309: Support software breakpoints in the debug server.: D20308: Add initial support for qSupported..
Sat, May 18, 6:32 PM
jhb updated subscribers of D20309: Support software breakpoints in the debug server..
Sat, May 18, 6:31 PM
jhb created D20309: Support software breakpoints in the debug server..
Sat, May 18, 6:29 PM
jhb created D20308: Add initial support for qSupported..
Sat, May 18, 6:27 PM
jhb created D20307: Add support for writing to guest memory in the debug server..
Sat, May 18, 6:27 PM

Fri, May 10

jhb accepted D20230: Remove obsolete 10 and 10/100 Ethernet NIC drivers per fcp-0101..

Just don't forget 'Relnotes: yes' on each commit? (I assume you were planning to do that but didn't see it in my spot check of a GitHub commit log)

Fri, May 10, 8:30 PM
jhb reopened D20117: Restructure mbuf send tags to provide stronger guarantees..
Fri, May 10, 8:16 PM
jhb closed D20117: Restructure mbuf send tags to provide stronger guarantees..
Fri, May 10, 8:15 PM
jhb committed rS347465: Apply r280991 to ip6_fragment..
Apply r280991 to ip6_fragment.
Fri, May 10, 8:15 PM
jhb added a comment to D20117: Restructure mbuf send tags to provide stronger guarantees..

Whoops, missed bz's earlier note about ip6_fragment, so I will pull that out and merge it separately.

Fri, May 10, 7:57 PM
jhb added a comment to D20117: Restructure mbuf send tags to provide stronger guarantees..

@bz any thoughts on the ip6_fragment change?

Fri, May 10, 7:55 PM
jhb updated the diff for D20117: Restructure mbuf send tags to provide stronger guarantees..
  • Rebase.
  • Remove checks that are currently always true.
  • Drop __inline.
Fri, May 10, 7:46 PM
jhb committed rS347443: Complete the MFC of 345963 after the MFC of 333174..
Complete the MFC of 345963 after the MFC of 333174.
Fri, May 10, 4:51 PM
jhb committed rS347437: MFC 342944: Fix spelling of identifier.
MFC 342944: Fix spelling of identifier
Fri, May 10, 4:38 PM
jhb committed rS347436: MFC 340707:.
MFC 340707:
Fri, May 10, 4:36 PM
jhb committed rS347435: MFC 338957:.
MFC 338957:
Fri, May 10, 4:33 PM

Thu, May 9

jhb committed rS347416: MFC 332479: Add SDT probes to vmexit on Intel..
MFC 332479: Add SDT probes to vmexit on Intel.
Thu, May 9, 11:57 PM
jhb committed rS347412: MFC 334272: bhyve: guarantee NUL termination.
MFC 334272: bhyve: guarantee NUL termination
Thu, May 9, 10:42 PM
jhb committed rS347411: MFC 334271: bhyve: fix small memory leak in virtio console.
MFC 334271: bhyve: fix small memory leak in virtio console
Thu, May 9, 10:39 PM
jhb committed rS347409: MFC 333639:.
MFC 333639:
Thu, May 9, 10:32 PM
jhb committed rS347406: MFC 333235:.
MFC 333235:
Thu, May 9, 8:30 PM

Wed, May 8

jhb committed rS347374: MFC 333174: Use PCI power-mgmt to reset a device if FLR fails..
MFC 333174: Use PCI power-mgmt to reset a device if FLR fails.
Wed, May 8, 11:24 PM
jhb accepted D20163: Remove IPSEC from GENERIC due to performance issues.
Wed, May 8, 8:11 PM
jhb added a comment to D20163: Remove IPSEC from GENERIC due to performance issues.
In D20163#434567, @jhb wrote:

FWIW, my limited testing of IPsec doesn't use if_ipsec, but instead I used setkey. I think having the rc.d scripts for 'ipsec_enable' autoloading ipsec.ko is reasonable. Maybe if there are ports scripts for IKE daemons (raccoon, strongswan?) those rc.d scripts should also try to load modules they need.

Now that @ae has added this feature to autoload ipsec.ko, are you OK with this change?

Wed, May 8, 8:11 PM